CA2981821A1 - Novel antibodies inhibiting c-met dimerization, and uses thereof - Google Patents

Abstract

The present invention relates to a process for the selection of anti c-Met antibodies capable to inhibit both ligand-dependent and ligand-independent activation of c-Met. More particularly, said process is based on the inhibition of the c-Met dimerization. In another aspect, the present invention concerns such antibodies and compositions comprising such antibodies for the preparation of a medicament to treat cancer. Diagnosis process and kits are also part of the invention.

Description

NOVEL ANTIBODIES INHIBITING c-MET DIMERIZATION, AND USES THEREOF
TECHNICAL FIELD
The present invention relates to novel antibodies capable of binding specifically to the human c-Met receptor and/or capable of specifically inhibiting the tyrosine kinasc activity of said receptor, especially monoclonal antibodies of murine, chimeric and humanized origin, as well as the amino acid and nucleic acid sequences coding for these antibodies. More particularly, antibodies according to the invention are capable of 1 0 inhibiting the c-Met dimerization. The invention likewise comprises the use of these antibodies as a medicament for the prophylactic and/or therapeutic treatment of cancers or any pathology connected with the overexpression of said receptor as well as in processes or kits for diagnosis of illnesses connected with the overexpression of c-Met.
The invention finally comprises products and/or compositions comprising such antibodies in combination with other antibodies and/or chemical compounds directed against other growth factors involved in tumor progression or metastasis and/or compounds and/or anti-cancer agents or agents conjugated with toxins and their use for the prevention and/or the treatment of certain cancers.
BACKGROUND OF THE INVENTION
Receptor tyrosine kinase (RTK) targeted agents such as trastuzumab, cetuxirnab, bevacizumab, imatinib and gefitinib inhibitors have illustrated the interest of targeting this protein class for treatment of selected cancers.
c-Met, is the prototypic member of a sub-family of RTKs which also includes RON and SEA. The c-Met RTK family is structurally different from other RTK
families and is the only known high-affinity receptor for hepatocytc growth factor (HGF), also called seater factor (SF) D.P. Bottaro et al., Science 1991, 251: 802-804; L.
Naldini et al., Eur. Mol. Biol. Org. J. 1991, 10:2867-2878]. c-Met and HGF are widely expressed in a variety of tissue and their expression is normally restricted to cells of epithelial and mesenchymal origin respectively [M.F. Di Renzo et al., Oncogene 1991, 6:1997-2003;
E. Sonnenberg et al., J. Cell. Biol. 1993, 123:223-235]. They arc both required for normal mammalian development and have been shown to be particularly important in cell migration, morphogenic differentiation, and organization of the three-dimensional tubular structures as well as growth and angiogcnesis [F. Baldt et al., Nature 1995, 376:768-771; C. Schmidt et al., Nature. 1995:373:699-702; Tsarfaty et at., Science 1994, 263:98-1011. While the controlled regulation of c-Met and HGF have been shown to be important in mammalian development, tissue maintenance and repair [Nagayama T, Nagayama M, Kohara S, Kamiguchi H, Shibuya M, Katoh Y, Itoh J, Shinohara Y., Brain Res. 2004, 5;999(2):155-66; Tahara Y, Id A, Yamamoto S, Miyata Y, Uto H, Boni T, IIayashi K, Tsubouchi II., .1 Pharmacol Exp Then 2003, 307(1):146-51], their dysregulation is implicated in the progression of cancers.
Aberrant signalling driven by inappropriate activation of c-Met is one of the most frequent alteration observed in human cancers and plays a crucial role in tumorigenesis and metastasis [Birchmeier et al., Nat. Rev. Mol. Cell Biol.
2003, 4:915-925; L. Trusolino and Comoglio P. M., Nat Rev. Cancer. 2002, 2(4):289-300].
Inappropriate c-Met activation can arise by ligand-dependent and independent mechanisms, which include overcxpression of c-Met, and/or paracrine or autocrine activation, or through gain in function mutation [J.G. Christensen, Burrows J.
and Salgia R., Cancer Latters. 2005, 226:1-26]. However an oligomerization of c-Met receptor, in presence or in absence of the ligand, is required to regulate the binding affinity and binding kinetics of the kinase toward ATP and tyrosine-containing peptide substrates [Hays JL, Watowich SJ, Biochemistry, 2004 Aug 17, 43:10570-8].
Activated c-Met recruits signalling effectors to its multidocking site located in the cytoplasm domain, resulting in the activation of several key signalling pathways, including Ras-MAPK, PI3K, Src and Stat3 [Gao CF, Vande Woude GF, Cell Res. 2005, 15(1):49-51;
Furgc KA, Zhang YW, Vande Woude GF, Oncogene. 2000, 19(49):5582-9]. These pathways are essential for tumour cell proliferation, invasion and angiogenesis and for evading apoptosis [Furge KA, Zhang YW, Vande Woude GF, Oncogene, 2000, 19(49):5582-9; Gu H, Neel BG, Trends Cell Biol. 2003 Mar, 13(3):122-30; Fan S, Ma YX, Wang JA, Yuan RQ, Meng Q, Cao Y, Laterra JJ, Goldberg ID, Rosen EM, Oncogene. 2000 Apr 27, 19(18):2212-23]. In addition, a unique facet of the c-Met signalling relative to other RTK is its reported interaction with focal adhesion complexes and non kinasc binding partners such as a6134 integrins [Trusolino L, Bertotti A, Comoglio PM, Cell. 2001, 107:643-54], CD44v6 [Van der Voort R, Taher TE, Wielenga VJ, Spaargaren M, Prevo R, Smit L, David G, Hartmann G, Gherardi E, Pals ST, J Biol Chem. 1999, 274(10):6499-506], Plexin B1 or semaphorins [Giordano S, Corso S, Conrotto P, Artigiani S, Gilestro G, Barberis D, Tarnagnone L, Comoglio PM, Nat Cell Biol. 2002, 4(9):720-4; Conrotto P, Valdembri D, Corso S, Serini G, Tamagnone L, Comoglio PM, Bussolino F, Giordano S, Blood. 2005, 105(11):4321-9;
Conrotto P. Corso S, Gambcrini S, Comoglio PM, Giordano S, Oncogene. 2004, 23:5131-7] which may further add to the complexity of regulation of cell function by this receptor. Finally recent data demonstrate that c-Met could be involved in tumor resistance to gefitinib or erlotinib suggesting that combination of compound targeting both EGFR and c-Met might be of significant interest [Engelman JA at al., Science, 2007, 316:1039-43].
In the past few years, many different strategies have been developed to attenuate c-Met signalling in cancer cell lines. These strategies include i) neutralizing antibodies against c-Met or HCiF/SF [Cao B, Su Y, Oskarsson M, Zhao P, Kort EJ, Fisher RJ, Wang LM, Vande Woude GF, Proc Natl Acad Sci U S A. 2001, 98(13):7443-8;
Martens T, Schmidt NO, Eckerich C, Fillbrandt R, Merchant M, Schwall R, Westphal M, Lamszus K, Clin Cancer Res. 2006, 12(20):6144-52] or the use of HGF/SF
antagonist NK4 to prevent ligand binding to c-Met [Kuba K, Matsumoto K, Date K, Shimura H, Tanaka M, Nakamura T, Cancer Res., 2000, 60:6737-43], ii) small ATP

2 0 binding site inhibitors to c-Met that block kinase activity [Christensen JG, Schreck R, Burrows J, Kuruganti P. Chan E, Le P, Chen J, Wang X, Ruslim L, Blake R, Lipson KE, Ramphal J, Do S, Cui JJ, Cherrington JM, Mendel DB, Cancer Res. 2003, 63:7345-55], iii) engineered SH2 domain polypeptide that interferes with access to the multidocking site and RNAi or ribozyme that reduce receptor or ligand expression.
Most of these approaches display a selective inhibition of c-Met resulting in tumor inhibition and showing that c-Met could be of interest for therapeutic intervention in cancer.
Within the molecules generated for c-Met targeting, some are antibodies.
The most extensively described is the anti-c-Met 5D5 antibody generated by Genentech [W096/38557] which behaves as a potent agonist when added alone in various models and as an antagonist when used as a Fab fragment. A monovalent engineered form of this antibody described as one armed 5D5 (0A5D5) and produced as a recombinant protein in E. Coll is also the subject of a patent application [W02006/015371] by Genentech. However, this molecule that could not be considered as an antibody because of its particular scarfold, displays also mutations that could be immunogenic in humans. In terms of activity, this unglycosylated molecule is devoided of effector functions and finally, no clear data demonstrate that 0A5D5 inhibits dimerization of c-Met. Moreover, when tested in the G55 in vivo model, a glioblastoma cell line that expresses c-Met but not HGF mRNA and protein and that grows independently of the ligand, the one armed anti-c-Met had no significant effect on G55 tumor growth suggesting that 0A5D5 acts primarily by blocking HGF binding and is not able to target tumors activated independently of I IGF [Martens T. et al, Clin. Cancer Res., 2006, 12(20):6144-6152].
Another antibody targeting c-Met is described by Pfizer as an antibody acting "predominantly as c-Met antagonist, and in some instance as a c-Met agonist"
[WO
2005/016382]. No data showing any effect of Pfizer antibodies on c-Met dimerization is described in this application.
One of the innovant aspects of the present invention is to generate mouse monoclonal antibodies without intrinsic agonist activity and inhibiting c-Met dimerization. In addition of targeting ligand-dependent tumors, this approach will also impair ligand-independent activations of c-Met due to its overexpression or mutations of the intra cellular domains which remained dependent to oligomerizat ion for signalling. Another aspect of the activity of such antibodies could be a steric hindrance for c-Met interaction with its partners that will result in impairment of c-Met functions.
These antibodies will be humanized and engineered preferentially, but not limited, as human IgG1 to get effector functions such as ADCC and CDC in addition to functions linked to the specific blockade of the c-Met receptor.
DISCLOSURE OF THE INVENTION
Surprisingly, for the first time, inventors have managed to generate an antibody capable of binding to c-Met but also capable of inhibiting the c-Met dimerization. If it is true that, in the prior art, it is sometimes suggested that an antibody capable of inhibiting the dimerization of c-Met with its partners could be an interesting one, it has never been disclosed, or clearly suggested, an antibody capable of doing so.
Moreover, regarding antibody specificity, it was not evident at all to succeed in the generation of such an active antibody.
In a first aspect, a subject of the present invention is a process for the generation and the selection of antibodies according to the invention.
5 More particularly, the invention concerns a process for the selection of an anti c-Met antibody, or one of its functional fragments or derivatives, capable to inhibit both ligand-dependent and ligand-independent activation of c-Met, said process comprising the following steps:
I) screening the generated antibodies and selecting antibodies capable to bind specifically to c-Met;
ii) evaluating in vitro the selected antibodies of step i) and selecting antibodies capable to inhibit at least 50 %, preferably at least 60 %, 70 % or 80 % of tumoral cell proliferation for at least one tumor type; and then iii) testing the selected antibodies of step ii) and selecting antibodies capable to inhibit the c-Met dimerizat ion.
As it was explained before, the inhibition of the c-Met dimerization is a capital aspect of the invention as such antibodies will present a real interest for a larger population of patients. Not only ligand-dependent activated c-Met cancer, as it was the case until the present invention, but also ligand-independent activated c-Met cancer could be treated with antibodies generated by the process of the present invention.
The generation of the antibody can be realized by any method known by the man skilled in the art, such as for example, fusion of a mycloma cell with spleen cells from immunized mice or other species compatible with the selected myeloma cells [Kohler & Milstein, 1975, Nature, 256:495-497]. The immunized animals could include transgcnic mice with human irnmunoglobulin loci which then directly produce human antibodies. Another possible embodiment could consist in using phage display technologies to screen libraries.
The screening step i) can be realized by any method or process known by the man skilled in the art. As non limitative examples, can be mentioned ELISA, BIAeore, immunohistochemistry, FACS analysis and functional screens. A preferred process consists in a screen by ELISA on the c-Met recombinant protein and then by FACS
analysis on at least a tumoral cell line to be sure that the produced antibodies will be able to also recognize the native receptor on tumor cells. This process will be described more precisely in the following examples.
In the same way, the step ii) can also be realized classically by known method or process such as, for example, using 3H-thymidine or any other DNA staining agent, MTT, ATP evaluation, etc. A preferred tumor cell model in the present invention can consist in the BxPC3 model.
By inhibiting c-Met dimerization, it must be understood preferably the c-Met homodimerization.
In a preferred embodiment of the step iii) of selection of the process of the invention, said step iii) consists in evaluating antibodies by BRET analysis on cells expressing both e-Met-RLue/c-Met-YFP and selecting antibodies capable to inhibit at least 30 %, preferably 35 %, 40 %, 45 %, 50 %, 55 % and most preferably 60 %
of the BRET signal.
The technology I3RET is a technology known as being representative of the protein dimcrization [Angers et al., PNAS, 2000, 97:3684-891.
The technology BRET, used in the step iii) of the process, is well known by the man skill in the art and will be detailed in the following examples. More particularly, BRET (Bioluminescence Resonance Energy Transfer) is a non-radiative energy transfer occurring between a bioluminescent donor (Renilla Luciferase (Rluc)) and a fluorescent acceptor, a mutant of GFP (Green Fluorescent Protein) or YFP (Yellow fluorescent protein). In the present case EYFP (Enhanced Yellow Fluorescent Protein) was used.
The efficacy of transfer depends on the orientation and the distance between the donor and the acceptor. Then, the energy transfer can occur only if the two molecules arc in close proximity (1-10 nm). This property is used to generate protein-protein interaction assays. Indeed, in order to study the interaction between two partners, the first one is genetically fused to the Renilla Lucifcrase and the second one to the yellow mutant of the GFP. Fusion proteins are generally, but not obligatory, expressed in mammalian cells. In presence of its membrane permeable substrate (coelenterazine), Rluc emits blue light. If the GFP mutant is closer than 10 nm from the Rluc, an energy transfer can occur and an additional yellow signal can be detected. The BRET signal is measured as the ratio between the light emitted by the acceptor and the light emitted by the donor. So the BRET signal will increase as the two fusion proteins are brought into proximity or if a conformational change bring RItic and GFP mutant closer.
If the BRET analysis consists in a preferred embodiment, any method known by the man skilled in the art can be used to measure c-Met dimerization. Without limitation, the following technologies can be mentioned: FRET (Fluorescence Resonance Energy Transfer), HTRF (Homogenous Time resolved Fluorescence), FLIM
(Fluorescence Lifetime Imaging Microscopy) or SW-FCCS single wavelength fluorescence cross-correlation spectroscopy).
Other classical technologies could also be used, such as Co-immunoprecipitation, Alpha screen, Chemical cross-linking, Double-Hybrid, Affinity Chromatography, ELISA or Far western blot.
In a second aspect, a subject of the invention is an isolated antibody, or one of its functional fragments or derivatives, being obtained by said process. Said antibody or one of its said fragments or derivatives, is capable of binding specifically to the human c-Met and, if necessary, preferably moreover capable of inhibiting the natural attachment of its ligand HGF and/or capable of specifically inhibiting the tyrosine kinase activity of said c-Met, said antibody being also capable to inhib c-Met dimerization. More particularly, said antibodies will be capable of inhibiting both ligand-dependent and ligand-independent activation of c-Met.
The expressions "functional fragments and derivatives" will be defined in details later in the present specification.
It must be understood here that the invention does not relate to the antibodies in natural form, that is to say they are not in their natural environment but that they have been able to be isolated or obtained by purification from natural sources, or else obtained by genetic recombination, or by chemical synthesis, and that they can then contain unnatural amino acids as will be described further on.
More particularly, according to another aspect of the invention, it is claimed an antibody, or one of its functional fragments or derivatives, said antibody being characterized in that it comprises at least one complementary determining region CDR
chosen from CDRs comprising the amino acid sequence SEQ ID Nos. 1 to 17 and 56 to 61.
Any antibody, or fragments or derivatives, having at least one CDR whose sequence has at least 80 % identity, preferably 85 %, 90 %, 95 % and 98 %
identity, after optimum alignment with the sequences SEQ ID Nos. 1 to 17 and 56 to 61 must be understood as a equivalent and, as a consequence, as being part of the invention.
By CDR regions or CDR(s), it is intended to indicate the hypervariable regions of the heavy and light chains of the immunoglobulins as defined by IMGT.
The 1MGT unique numbering has been defined to compare the variable domains whatever the antigen receptor, the chain type, or the species [Lefranc M.-P., Immunology Today 18, 509 (1997) / Lcfranc M.-P., The Immunologist, 7, 132-136 (1999)! Lefranc, M.-P., Pommie, C., Ruiz, M., Giudicelli, V., Foulquier, E., Truong, L, Thouvenin-Contet, V. and Lefranc, Dev. Comp. Immunol., 27, 55-77 (2003)]. In the IMGT unique numbering, the conserved amino acids always have the same position, for instance cysteine 23 (lst-CYS), tryptophan 41 (CONSERVED-TRP), hydrophobic amino acid 89, cysteine 104 (2nd-CYS), phenylalanine or tryptophan 118 (J-PHE
or J-TRP). The 1MGT unique numbering provides a standardized delimitation of the framework regions (FR1-IMG1': positions 1 to 26, FR2-IMGT: 39 to 55, FR3-1MG1':
66 to 104 and FR4-IMGT: 118 to 128) and of the complcmcntarity determining regions:
CDR1-IMGT: 27 to 38, CDR2-1MGT: 56 to 65 and CDR3-IMGT: 105 to 117. As gaps represent unoccupied positions, the CDR-IMGT lengths (shown between brackets and separated by dots, e.g. [8.8.13]) become crucial information. The IMGT unique numbering is used in 2D graphical representations, designated as IMGT Colliers de Perles [Ruiz, M. and Lefranc, M.-P., lmmunogenetics, 53, 857-883 (2002) /
Kaas, Q.
and Lefranc, M.-P., Current Bioinformatics, 2, 21-30 (2007)], and in 3D
structures in IMGT/3Dstructure-DB [Kaas, Q., Ruiz, M. and Lefranc, M.-P., T cell receptor and MHC structural data. Nucl. Acids. Res., 32, D208-D210 (2004)].
Three heavy chain CDRs and 3 light chain CDRs exist. The term CDR or CDRs is used here in order to indicate, according to the case, one of these regions or several, or even the whole, of these regions which contain the majority of the amino acid residues responsible for the binding by affinity of the antibody for the antigen or the epitope which it recognizes.
By "percentage of identity" between two nucleic acid or amino acid sequences in the sense of the present invention, it is intended to indicate a percentage of nucleotides or of identical amino acid residues between the two sequences to be compared, obtained after the best alignment (optimum alignment), this percentage being purely statistical and the differences between the two sequences being distributed randomly and over their entire length. The comparisons of sequences between two nucleic acid or amino acid sequences are traditionally carried out by comparing these sequences after having aligned them in an optimum manner, said comparison being able to be carried out by segment or by "comparison window". The optimum alignment of the sequences for the comparison can be carried out, in addition to manually, by means of the local homology algorithm of Smith and Waterman (1981) [Ad. App. Math.
2:482], by means of the local homology algorithm of Neddlema.n and Wunsch (1970) [J.
IVIol. Biol. 48: 443], by means of the similarity search method of Pearson and Lipman (1988) [Proc. Natl. Acad. Sci. USA 85:2444), by means of computer software using these algorithms (GAP, BESTF1T, FASTA and TFASTA in the Wisconsin Genetics Software Package, Genetics Computer Group, 575 Science Dr., Madison, WI, or else by BLAST N or BLAST P comparison software).
the percentage of identity between two nucleic acid or amino acid sequences is determined by comparing these two sequences aligned in an optimum manner and in which the nucleic acid or amino acid sequence to be compared can comprise additions or deletions with respect to the reference sequence for an optimum alignment between these two sequences. The percentage of identity is calculated by determining the number of identical positions for which the nucleotide or the amino acid residue is identical between the two sequences, by dividing this number of identical positions by the total number of positions in the comparison window and by multiplying the result obtained by 100 in order to obtain the percentage of identity between these two sequences.
For example, it is possible to use the BLAST program, "BLAST 2 sequences"
(Tatusova et al., "Blast 2 sequences - a new tool for comparing protein and nucleotide sequences", FEMS Microbiol Lett. 174:247-250) available on the site http://www.nebi.nlm.nih.gov/ gorf/b12.html, the parameters used being those given by default (in particular for the parameters "open gap penalty": 5, and "extension gap penalty": 2; the matrix chosen being, for example, the matrix "BLOSUM 62"
proposed by the program), the percentage of identity between the two sequences to be compared being calculated directly by the program.

WO 2009/007=127 PCT/EP2008/059026 By amino acid sequence having at least 80 %, preferably 85 %, 90 %, 95 % and 98 % identity with a reference amino acid sequence, those having, with respect to the reference sequence, certain modifications, in particular a deletion, addition or substitution of at least one amino acid, a truncation or an elongation are preferred. In the 5 case of a substitution of one or more consecutive or nonconsecutive amino acid(s), the substitutions arc preferred in which the substituted amino acids are replaced by "equivalent" amino acids. The expression "equivalent amino acids" is aimed here at indicating any amino acid capable of being substituted with one of the amino acids of the base structure without, however, essentially modifying the biological activities of 10 the corresponding antibodies and such as will be defined later, especially in the examples. These equivalent amino acids can be determined either by relying on their structural homology with the amino acids which they replace, or on results of comparative trials of biological activity between the different antibodies capable of being carried out.
By way of example, mention is made of the possibilities of substitution capable of being carried out without resulting in a profound modification of the biological activity of the corresponding modified antibody.
As non limitative example, the following table 1 is giving substitution possibilities conceivable with a conservation of the biological activity of the modified antibody. The reverse substitutions are also, of course, possible in the same conditions.

Table 1 Original residu Substitution(s) Ala (A) Val, Gly, Pro Arg (R) Lys, His Asti (N) Gin Asp (D) Glu Cys (C) Ser Gin (Q) Asn Glu (G) Asp Gly (G) Ala His (H) Arg Ile (I) Leu Leu (L) Ile, Val, Met Lys (K) Arg Met (M) Leu Phe (F) Tyr Pro (P) Ala Ser (S) Thr, Cys Thr (T) Ser Trp (W) Tyr Tyr (Y) Phe, Trp Val (V) Lett, Ala It must be understood here that the invention does not relate to the antibodies in natural form, that is to say they are not in their natural environment but that they have been able to be isolated or obtained by purification from natural sources, or else obtained by genetic recombination, or by chemical synthesis, and that they can then contain unnatural amino acids as will be described further on.
According a first approach, the antibody will be defined by its heavy chain 1 0 sequence. More particularly, the antibody of the invention, or one of its functional fragments or derivatives, is characterized in that it comprises a heavy chain comprising at least one CDR chosen from CDRs comprising the amino acid sequences SEQ Ill Nos. 1 to 9 and 56 to 58.
The mentioned sequences are the following ones:
SEQ ID No. 1: GYIFTAYT
SEQ ID No. 2: IKPNNGLA
SEQ ID No. 3: ARSEITTEFDY

SEQ ID No. 4: GYSFTDYT
SEQ ID No. 5: INPYNGGT
SEQ ID No. 6: AREEITKDFDF
SEQ ID No. 7: GYTFTDYN
SEQ ID No. 8: INPNNGGT
SEQ ID No. 9: ARGRYVGYYYAMDY
SEQ ID No. 56: GYTFTSYW
SEQ ID No. 57: INPTTGST
SEQ ID No. 58: AIGGYGSWFAY
The CDRs of the heavy chain could be chosen randomly in the previous sequences, i.e. SEQ ID Nos. 1 to 9 and 56 to 58.
According to a preferred aspect, the antibody of the invention, or one of its functional fragments or derivatives, comprises a heavy chain comprising at least one CDR chosen from CDR-H1, CDR-H2 and CDR-H3, wherein:
- CDR-H1 comprises the amino acid sequence SEQ ID No. 1, 4, 7 or 56, - CDR-H2 comprises the amino acid sequence SEQ ID No. 2, 5, 8 or 57, and - CDR-H3 comprises the amino acid sequence SEQ ID No. 3, 6, 9 or 58.
According to a first embodiment of said aspect, the antibody of the invention, or one of its functional fragments or derivatives, comprises a heavy chain comprising CDR-H1, CDR-H2 and CDR-H3, wherein CDR-HI comprises the amino acid sequence SEQ ID No. 1, CDR-H2 comprises the amino acid sequence SEQ ID No. 2 and CDR-H3 comprises the amino acid sequence SEQ ID No. 3.
More particularly, said antibody, or one of its functional fragments or derivatives, according to this first embodiment comprises a heavy chain of sequence comprising the amino acid sequence SEQ ID No. 18.
SEQ ID No. 18: EVQLQQSGPELVKPGASVK1SCKTSGYIFTAYTMHWVRQSLG
ESLDWIGGIKPNNGLANYNQKFKGKATLTVDKSSSTAYMDLRSLTSEDSAVYY
CARSEITTEFDYWGQGTALTVSS
According to a second embodiment of said aspect, the antibody of the invention, or one of its functional fragments or derivatives, comprises a heavy chain comprising CDR-H1, CDR-H2 and CDR-H3, wherein CDR-H1 comprises the amino acid sequence SEQ ID No. 4, CDR-H2 comprises the amino acid sequence SEQ ID No. 5 and CDR-H3 comprises the amino acid sequence SEQ ID No. 6.
The antibody, or one of its functional fragments or derivatives, according to said second embodiment will preferably comprise a heavy chain of sequence comprising the amino acid sequence SEQ ID No. 19.
SEQ ID No. 19: EVQLQQSGPELVKPGASMKISCKASGYSFTDYTLNWVKQSH
GKTLEWIGLINPYNGGTTYNQKFKGKATLTVDKSSSTAYMELLSLTSEDSAVY
YCAREEITKDFDFWGQGTTLTVSS
According to a third embodiment of said aspect, the antibody of the invention, or one of its functional fragments or derivatives, comprises a heavy chain comprising CDR-H1, CDR-H2 and CDR-H3, wherein CDR-H1 comprises the amino acid sequence SEQ ID No. 7, CDR-H2 comprises the amino acid sequence SEQ ID No. 8 and CUR-H3 comprises the amino acid sequence SEQ ID No. 9.
The antibody, or one of its functional fragments or derivatives, according to said third embodiment will preferably comprise a heavy chain of sequence comprising the amino acid sequence SEQ ID No. 20.
SEQ ID No. 20: EVLLQQSGPELVKPGASVKIPCKASGYTFTDYNIADWVKQSH
GMSLEWIGDINPNNGGTIFNQKFKGKATLTVDKSSSTAYMELRSLTSEDTAVYY
CARGRYVGYYYAMDYWGQGTSVTVSS
According to a fourth embodiment of said aspect, the antibody of the invention, or one of its functional fragments or derivatives, comprises a heavy chain comprising CDR-H1, CDR-H2 and CDR-I-13, wherein CDR-H1 comprises the amino acid sequence SEQ ID No. 56, CDR-H2 comprises the amino acid sequence SEQ ID No. 57 and CDR-H3 comprises the amino acid sequence SEQ ID No. 58.
The antibody, or one of its functional fragments or derivatives, according to said fourth embodiment will preferably comprise a heavy chain of sequence comprising the amino acid sequence SEQ ID No. 62.
SEQ ID No. 62:
QVQLQQSGAELAKPGASVKMSCKASGYTFTSYWMNWVKQRPGQGLEWIGYI
NPTTGSTDYNQKLKDKATLTADKSSNTAYMQLSSLTSEDSAVYYCAIGGYGSW
FAYWGQGTLVTVSA

In a second approach, the antibody will be now define by its light chain sequence. More particularly, according to a second particular aspect of the invention, the antibody, or one of its functional fragments or derivatives, is characterized in that it comprises a light chain comprising at least one CDR chosen from CDRs comprising the amino acid sequence SEQ -ID Nos. 10 to 17 and 59 to 61.
The mentioned sequences arc the following ones:
SEQ ID No. 10: ESVDSYANSF -SEQ ID No. 11: RAS
SEQ ID No. 12: QQSKEDPLT
SEQ ID No. 13: ESIDTYGNSF
SEQ ID No. 14: QQSNEDPFT
SEQ ID No. 15: ENlYSN
SEQ Ill No. 16: AM' SEQ Ill No. 17: QHFWGPPYT
SEQ ID No. 59: SSVSSTY
SEQ ID No. 60: TTS
SEQ ID No. 61: HQWSSYPFT
The CDRs of the light chain could be chosen randomly in the previous sequences, i.e. SEQ ID Nos. 10 to 17 and 59 to 61.
According to another preferred aspect, the antibody of the invention, or one of its functional fragments or derivatives, comprises a light chain comprising at least one CDR chosen from CDR-L1, CDR-L2 and CDR-L3, wherein:
- CDR-L1 comprises the amino acid sequence SEQ ID No. 10, 13, 15 or 59, - CDR-L2 comprises the amino acid sequence SEQ ID No. 11, 16 or 60, and - CDR-L3 comprises the amino acid sequence SEQ ID No. 12, 14, 17 or 61.
According to a first embodiment of said another aspect, the antibody of the invention, or one of its functional fragments or derivatives, comprises a light chain comprising CDR-L1, CDR-L2 and CDR-L3, wherein CDR-L1 comprises the amino acid sequence SEQ ID No. 10, CDR-L2 comprises the amino acid sequence SEQ ID
No. 11 and CDR-L3 comprises the amino acid sequence SEQ ID No. 12.

=
More particularly, said antibody, or one of its functional fragments or derivatives, according to this first embodiment comprises a light chain of sequence comprising the amino acid sequence SEQ ID No. 21.
SEQ ID No. 21: DIVLTQSPASLAVSLGQRATISCRASESVDSYANSFMHWYQQ

DPLTFGSGTKLEMK
According to a second embodiment of said another aspect, the antibody of the invention, or one of its functional fragments or derivatives, comprises a light chain 10 comprising CDR-L1, CDR-1L2 and CDR-L3, wherein CDR-L1 comprises the amino acid sequence SEQ ID No. 13, CDR-L2 comprises the amino acid sequence SEQ ID
No. 11 and CDR-L3 comprises the amino acid sequence SEQ Ill No. 14.
The antibody, or one of its functional fragments or derivatives, according to said second embodiment will preferably comprise a light chain of sequence comprising the 15 amino acid sequence SEQ ID No. 22.
SEQ ID No. 22: GIVLTQSPASLAVSLGQRATISCRVSESIDTYGNSFIHWYQQKP
GQPPKWYRASNLESGIPARFSGSGSRTDFTLTINPVEADDSATYYCQQSNEDPF
TFGSGTKLEMK
According to a third embodiment of said another aspect, the antibody of the invention, or one of its functional fragments or derivatives, comprises a light chain comprising CDR-L1, CDR-L2 and CDR-L3, wherein CDR-L1 comprises the amino acid sequence SEQ ID No. 15, CDR-L2 comprises the amino acid sequence SEQ ID
No. 16 and CDR-L3 comprises the amino acid sequence SEQ ID No. 17.
The antibody, or one of its functional fragments or derivatives, according to said third embodiment will preferably comprise a light chain of sequence comprising the amino acid sequence SEQ ID No. 23.
SEQ ID No. 23: DIQMTQSPASLSVSVGETVTITCRASENIYSNLAWYQQKQGKSP
QLLVYAATNLVDGVPSRFSGSGSGTQYSLKINSLQSEDEGSYYCQHFWGPPYTE
GGGTKLEIK

According to a fourth embodiment of said another aspect, the antibody of the invention, or one of its functional fragments or derivatives, comprises a light chain comprising CDR-L1, CDR-L2 and CDR-L3, wherein CDR-L1 comprises the amino acid sequence SEQ ID No. 59, CDR-L2 comprises the amino acid sequence SEQ ID
No. 60 and CDR-L3 comprises the amino acid sequence SEQ ID No. 61.
The antibody, or one of its functional fragments or derivatives, according to said third embodiment will preferably comprise a light chain of sequence comprising the amino acid sequence SEQ ID No. 63.
SEQ ID No. 63:
QIVLTQSPAIMSASPGEKVTLTCSASSSVSSTYLYWYQQIUGSSPKLWIYTTSIL
ASGVPARFSGSGSGTSYSLTISSMETEDAASYECIIQWSSYPFTFGSGTKLDIK
According a third approach, the antibody will be now defined both by its light chain sequence and its heavy chain sequence. The antibody of the invention, or one of its functional fragments or derivatives, is characterized in that it comprises a heavy chain comprising the amino acid sequence SEQ ID No. 18, 19, 20 or 62 and a light chain comprising the amino acid sequence SEQ ID No. 21, 22, 23 or 63.
More particularly, a preferred antibody, or one of its functional fragments or derivatives, according to the invention, named 224G11, comprises a heavy chain comprising CDR-H1, CDR-H2 and CDR-H3 comprising respectively the amino acid sequence SEQ ID Nos. 1, 2 and 3; and a light chain comprising CDR-L1, CDR-L2 and CDR-L3 comprising respectively the amino acid sequence SEQ ID Nos. 10, 11 and 12.
In another aspect, the antibody 224G11 comprises a heavy chain comprising the amino acid sequence SEQ ID No. 18 and a light chain comprising the amino acid sequence SEQ ID No. 21.
Another preferred antibody, or one of its functional fragments or derivatives, according to the invention, named 227H1, comprises a heavy chain comprising CDR-H1, CDR-H2 and CDR-H3 comprising respectively the amino acid sequence SEQ ID
Nos. 4, 5 and 6; and a light chain comprising CDR-L1, CDR-L2 and CDR-L3 comprising respectively the amino acid sequence SEQ ID Nos. 13, 11 and 14.

In another aspect, the antibody 227H1 comprises a heavy chain comprising the amino acid sequence SEQ ID No. 19 and a light chain comprising the amino acid sequence SEQ ID No. 22.
Still another preferred antibody, or one of its functional fragments or derivatives, named 223C4, comprises a heavy chain comprising CDR-H1, CDR-H2 and CDR-H3 comprising respectively the amino acid sequence SEQ ID Nos. 7, 8 and 9; and a light chain comprising CDR-L1, CDR-L2 and CDR-L3 comprising respectively the amino acid sequence SEQ ID Nos. 15, 16 and 17.
In another aspect, the antibody 223C4 comprises a heavy chain comprising the 1 0 amino acid sequence SEQ ID No. 20 and a light chain comprising the amino acid sequence SEQ ID No. 23.
Still another preferred antibody, or one of its functional fragments or derivatives, named 11E1, comprises a heavy chain comprising CDR-H1, CDR-H2 and CDR-H3 comprising respectively the amino acid sequence SEQ Ill Nos. 56, 57 and 58;
and a light chain comprising CDR-L1, CDR-L2 and CDR-L3 comprising respectively the amino acid sequence SEQ ID Nos. 59, 60 and 61.
In another aspect, the antibody 11E1 comprises a heavy chain comprising the amino acid sequence SEQ ID No. 62 and a light chain comprising the amino acid sequence SEQ ID No. 63.
According to another aspect, the invention relates to murine hybridoma capable of secreting monoclonal antibodies according to the present invention, especially hybridoma of murinc origin such as deposited at the Collection Nationale de Cultures de Microorganismes (CNCM, National Collection of Microorganism Cultures) (Institut Pasteur, Paris, France).
The monoclonal antibodies according to the invention, or one of their functional fragments or derivatives, are characterized in that said antibodies are secreted by the hybridoma deposited at the CNCM on 03/14/2007 under the numbers CNCM 1-3724 (corresponding to 11E1), 1-3731 (corresponding to 224G11), 1-3732 (corresponding to 227H1) and on 07/06/2007 under the number 1-3786 (corresponding to 223C4).
These hybridoma consist in murine hybridoma resulting in the cellular fusion of immunized mouse splenocytes with a myeloma cell line (Sp20 Ag14).
The following table 2 regroups elements concerning the preferred antibodies.

Table 2 Prot. Nucl. Prot; Nucl. Prot. Nucl. Prot.
Nucl.
SEQ IT) SEQ ID SEQ IT) SEQ ID SEQ ID SEQ ID SEQ ID SEQ ID

H. chain 18 41 19 49 20 43 62 70 L. chain 21 44 22 45 23 46 63 71 From table 2, it clearly appears that CDR-L2 of the antibodies 227H1 and 224011 is similar. This example clearly supports the claims of the present application covering antibodies comprising at least one CDR randomly chosen through described CDR sequences.
According to a preferred embodiment, the invention relates to monoclonal antibodies.
1 0 The term Monoclonal Antibody >> or is used in accordance with its ordinary meaning to denote an antibody obtained from a population of substantially homogeneous antibodies, i.e. the individual antibodies comprising the population are identical except for possible naturally occurring mutations that may be present in minor amounts. In other words, a monoclonal antibody consists in a homogenous antibody resulting from the proliferation of a single clone of cells (e.g., hybridoma cells, eukaryotic host cells transfected with DNA encoding the homogenous antibody, prokaryotic host cells transformed with DNA encoding the homogenous antibody, etc.), and which is generally characterized by heavy chains of a single class and subclass, and light chains of a single type. Monoclonal antibodies are highly specific, being directed against a single antigen. Furthermore, in contrast to poly-clonal antibodies preparations that typically include different antibodies directed against different determinants, or cpitope, each monoclonal antibody is directed against a single determinant on the antigen.

In the present description, the terms polypeptides, polypeptide sequences, amino acid sequences, peptides and proteins attached to antibody compounds or to their sequence are interchangeable.
According to a likewise particular aspect, the present invention relates to a chimeric antibody, or one of its functional fragments, according to the invention, characterized in that said antibody moreover comprises the light chain and heavy chain constant regions derived from an antibody of a species heterologous to the mouse, especially man, and in a preferred manner in that the light chain and heavy chain constant regions derived from a human antibody are respectively the kappa and gamma-1, gamma-2 or gamma-4 region.
In the present application, IgG I are preferred to get effector functions, and most preferably ADCC and CDC.
The skilled artisan will recognize that effector functions include, for example, CI q binding; complement dependent cytotoxicity (CDC); Fe receptor binding;
antibody-dependent cell-mediated cytotoxicity (ADCC); phagocytosis; and down regulation of cell surface receptors (e.g. B cell receptor; BCR).
The antibodies according to the present invention, arc preferably specific monoclonal antibodies, especially of murine, chimeric or humanized origin, which can be obtained according to the standard methods well known to the person skilled in the art.
In general, for the preparation of monoclonal antibodies or their functional fragments or derivatives, especially of murine origin, it is possible to refer to techniques which are described in particular in the manual "Antibodies" (Harlow and Lane, Antibodies: A Laboratory Manual, Cold Spring Harbor Laboratory, Cold Spring Harbor NY, pp. 726, 1988) or to the technique of preparation from hybridomas described by Kohler and Milstein (Nature, 256:495-497, 1975).
The monoclonal antibodies according to the invention can be obtained, for example, from an animal cell immunized against the c-Met, or one of its fragments containing the epitope specifically recognized by said monoclonal antibodies according to the invention. Said c-Met, or one of its said fragments, can especially be produced according to the usual working methods, by genetic recombination starting with a nucleic acid sequence contained in the cDNA sequence coding for the c-Met or by peptide synthesis starting from a sequence of amino acids comprised in the peptide sequence of the c-Met.
The monoclonal antibodies according to the invention call, for example, be purified on an affinity column on which the c-Met or one of its fragments containing the epitope specifically recognized by said monoclonal antibodies according to the invention has previously been immobilized. More particularly, said monoclonal antibodies can be purified by chromatography on protein A and/or G, followed or not followed by ion-exchange chromatography aimed at eliminating the residual protein contaminants as well as the DNA and the LPS, in itself followed or not thllowed by 1 0 exclusion chromatography on SepharoscTM gel in order to eliminate the potential aggregates due to the presence of dimcrs or of other multimers. In an even more preferred manner, the whole of these techniques can be used simultaneously or successively.
Chimeric or humanized antibodies are likewise included in antibodies according 1 5 to the present invention.
By chimeric antibody, it is intended to indicate an antibody which contains a natural variable (light chain and heavy chain) region derived from an antibody of a given species in combination with the light chain and heavy chain constant regions of an antibody of a species heterologous to said given species (e.g. mouse, horse, rabbit, 20 dog, cow, chicken, etc.).
The antibodies or their fragments of chimeric type according to the invention can be prepared by using the techniques of genetic recombination. For example, the chimeric antibody can be produced by cloning a recombinant DNA containing a promoter and a sequence coding for the variable region of a non-human, especially murine, monoclonal antibody according to the invention and a sequence coding for the constant region of human antibody. A chimeric antibody of the invention encoded by such a recombinant gene will be, for example, a mouse-man chimera, the specificity of this antibody being determined by the variable region derived from the murinc DNA
and its isotypc determined by the constant region derived from the human DNA.
For the methods of preparation of chimeric antibodies, it is possible, for example, to refer to the documents Verhoeyn et al. (BioEssays, 8:74, 1988), Morrison et al. (Proc.
Natl. Acad.
Sci. USA 82:6851-6855, 1984) ou le brevet US 4,816,567.

By humanized antibody, it is intended to indicate an antibody which contains CDR regions derived from an antibody of nonhuman origin, the other parts of the antibody molecule being derived from one (or from several) human antibodies.
Moreover, some of the residues of thc segments of the skeleton (called FR) can be modified in order to conserve the affinity of the binding (Jones et al., Nature, 321:522-525, 1986; Verhoeycn et al., Science, 239:1534-1536, 1988; Riechmann et al., Nature, 332:323-327, 1988).
The humanized antibodies according to the invention or their fragments can be prepared by techniques known to the person skilled in the art (such as, for example, those described in the documents Singer et at., J. lmmun. 150:2844-2857, 1992;
Mountain et al., Biotechnol. Genet. Eng. Rev., 10: 1-142, 1992; or Bebbington et al., Bio/Technology, 10:169-175, 1992).
Other humanization method are known by the man skill in the art as, for example, the -CDR Grafting" method described by Protein Design Lab (PDL) in the patent applications EP 0 451261, EP 0 682 040, EP 0 9127, EP 0 566 647 or US
5,530,101, US 6,180,370, US 5,585,089 and US 5,693,761. The following patent applications can also be mentioned: US 5,639,641; US 6,054,297; US 5,886,152 and US
5,877,293.
By "functional fragment" of an antibody according to the invention, it is 2 0 intended to indicate in particular an antibody fragment, such as Fv, scFy (Sc for single chain), Fab, F(ab')2. Fab', scFv-Fc fragments or diabodies, or any fragment of which the half-life time would have been increased by chemical modification, such as the addition of poly(alkylene) glycol such as poly(ethylene) glycol ("PEGylation") (pegylated fragments called Fv-PEG, scFv-PEG, Fab-PEG, F(ab')2-PEG or Fab'-PEG) ("PEG" for Poly(Ethylene) Glycol), or by incorporation in a liposome, said fragments having at least one of the characteristic CDRs of sequence SEQ ID Nos. Ito 17 and 56 to 61 according to the invention, and, especially, in that it is capable of exerting in a general manner an even partial activity of the antibody from which it is descended, such as in particular the capacity to recognize and to bind to the c-Met, and, if necessary, to inhibit the activity of the c-Met.
Preferably, said functional fragments will be constituted or will comprise a partial sequence of the heavy or light variable chain of the antibody from which they are derived, said partial sequence being sufficient to retain the same specificity of binding as the antibody from which it is descended and a sufficient affinity, preferably at least equal to 1/100, in a more preferred manner to at least 1/10, of that of the antibody from which it is descended, with respect to the c-Met. Such a functional fragment will contain at the minimum 5 amino acids, preferably 6,7, 8, 9, 10, 12, 15, 25, 50 and 100 consecutive amino acids of the sequence of the antibody from which it is descended.
Preferably, these functional fragments will be fragments of Fv, scFv, Fab, F(ab')2, F(ab'), scFv-Fc type or diabodies, which generally have the same specificity of binding as the antibody from which they are descended. In a more preferred embodiment of the invention, these fragments are selected among divalent fragments such as F(ab')2 fragments. According to the present invention, antibody fragments of the invention can be obtained starting from antibodies such as described above by methods such as digestion by enzymes, such as pepsin or papain and/or by cleavage of the disulfide bridges by chemical reduction. In another manner, the antibody fragments comprised in the present invention can be obtained by techniques of genetic recombination likewise well known to the person skilled in the art or else by peptide synthesis by means of, for example, automatic peptide synthesizers such as those supplied by the company Applied Biosystems, etc.
By "divalent fragment", it must be understood any antibody fragments comprising two arms and, more particularly, F(ab')2 fragments.
More particularly, the invention comprises the antibodies, or their functional fragments, according to the present invention, especially chimeric or humanized antibodies, obtained by genetic recombination or by chemical synthesis.
By derivatives >> of an antibody according to the invention, it is meant binding protein comprising a protein scaffold and at least on of the CDRs selected from the original antibody in order to maintain the binding capacity. Such compounds arc well known by the man skilled in the art and will be described in more details in the following specification.
More particularly, the antibody, or one of its functional fragments or derivatives, according to the invention is characterized in that sand derivative consists in a binding protein comprising a scaffold on which at least one CDR has been grafted for the conservation of the original antibody paratopic recognizing properties.

One or several sequences through the 6 CDR sequences described in the invention can be presented on a protein scaffold. In this case, the protein scaffold reproduces the protein backbone with appropriate folding of the grafted CDR(s), thus allowing it (or them) to maintain their antigen paratopic recognizing properties.
The man skilled in the art knows how to select the protein scaffold on which at least one CDR selected from the original antibody could be grafted. More particularly, it is known that, to be selected, such scaffold should display several features as follow (Skerra A., J. Mol. Recogn., 13, 2000, 167-187):
- phylogcnctically good conservation, - robust architecture with a well known three-dimensional molecular organization (such as, for example, crystallography or NMR), - small size, - no or only low degree of post-translational modifications, - easy to produce, express and purify.
Such protein scaffold can be, but without limitation, structure selected from the group consisting in fibroncctin and preferentially the tenth fibronectin type III domain (FNfn10), lipocalin, anticalin (Skerra A., J. Biotechnol., 2001, 74(4):257-75), the protein Z derivative from the domain B of staphylococcal protein A, thioredoxin A or any protein with repeated domain such as "ankyrin repeat" (Kohl et al., PNAS, 2003, vol. 100, No. 4, 1700-1705), "armadillo repeat", "leucin-rich repeat" or "tetratricopeptide repeat".
It could also be mentioned scaffold derivative from toxins (such as, for example, scorpion, insect, plant or mollusc toxins) or protein inhibitors of neuronal nitric oxydc synthasc (PIN).
As non limitative example of such hybrid constructions, it can be mentioned the insertion of the CDR-H1 (heavy chain) of an anti-CD4 antibody, i.e. the 13B8.2 antibody, into one of the exposed loop of the PIN. The binding properties of the obtained binding protein remain similar to the original antibody (Bes et al., BBRC 343, 2006, 334-344). It can also be mentioned the grafting of the CDR-H3 (heavy chain) of an anti-lyzozyme VHH antibody on a loop of neocarzinostatine (Nicaise et al., 2004).
In the case of the present invention, an interesting CDR to conserve could be, without limitation, the CDR-L2 as it is conserved in two identified antibodies of the invention, i.e. 227H1 and 224G11.
As above mentioned, such protein scaffold can comprise from 1 to 6 CDR(s) from the original antibody. In a preferred embodiment, but without any limitation, the man skilled in the art would select at least a CDR from the heavy chain, said heavy chain being known to be particularly implicated in the antibody specificity.
The selection of the CDR(s) of interest will be evident for the man of the art with known method (BES et al., FEBS letters 508, 2001, 67-74).
As an evidence, these examples are not limitative and any other scaffold known or described must be included in the present specification.
According to a novel aspect, the present invention relates to an isolated nucleic acid, characterized in that it is chosen from the following nucleic acids:
a) a nucleic acid, DNA or RNA, coding for an antibody, or one of its functional fragments or derivatives, according to the invention;
b) a nucleic acid comprising a DNA sequence selecting from the group of sequences consisting of:
- a nucleic sequence comprising the sequences SEQ ID No. 24, SEQ ID No. 25, SEQ ID No. 26 and the sequences SEQ ID No. 33, SEQ ID No. 34 and SEQ ID No.
35;
- a nucleic sequence comprising the sequences SEQ ID No. 27, SEQ ID No. 28, SEQ ID No. 29 and the sequences SEQ ID No, 36, SEQ ID No. 34 and SEQ ID No.
37;
- a nucleic sequence comprising the sequences SEQ ID No. 30, SEQ ID No. 31, SEQ ID No. 32 and the sequences SEQ ID No. 38, SEQ ID No. 39 and SEQ ID No.
40;
and - a nucleic sequence comprising the sequences SEQ ID No. 64, SEQ ID No. 65, SEQ ID No. 66 and the sequences SEQ ID No. 67, SEQ ID No. 68 and SEQ ID No.
69;
c) a nucleic acid comprising a DNA sequence selecting from the group of sequences consisting of:
- a nucleic sequence comprising the sequences SEQ ID No. 41 and SEQ ID No.
44;
- a nucleic sequence comprising the sequences SEQ ID No. 42 and SEQ ID No.
45;

- a nucleic sequence comprising the sequences SEQ ID No. 43 and SEQ ID No.
46, and - a nucleic sequence comprising the sequences SEQ ID No. 70 and SEQ ID No.
71;
5 d) the corresponding RNA nucleic acids of the nucleic acids as defined in b) or c);
c) the complementary nucleic acids of the nucleic acids as defined in a), b) and c); and f) a nucleic acid of at least 18 nucleotides capable of hybridizing under 10 conditions of hight stringency with at least one of the CDRs of sequence SEQ ID Nos.
24 to 40 and 64 to 69.
By nucleic acid, nucleic or nucleic acid sequence, polynucleotide, oligonucicotide, polynucleotide sequence, nucleotide sequence, terms which will be employed indifferently in the present invention, it is intended to indicate a precise 15 linkage of nucleotides, which arc modified or unmodified, allowing a fragment or a region of a nucleic acid to be defined, containing or not containing unnatural nucleotides, and being able to correspond just as well to a double-stranded DNA, a single-stranded DNA as to the transcription products of said DNAs.
It must also be understood here that the present invention does not concern the 20 nucleotide sequences in their natural chromosomal environment, that is to say in the natural state. It concerns sequences which have been isolated and/or purified, that is to say that they have been selected directly or indirectly, for example by copy, their environment having been at least partially modified. It is thus likewise intended to indicate here the isolated nucleic acids obtained by genetic recombination by means, for 25 example, of host cells or obtained by chemical synthesis.
An hybridization under conditions of high stringency signifies that the temperature conditions and ionic strength conditions are chosen in such a way that they allow the maintenance of the hybridization between two fragments of complementary DNA. By way of illustration, conditions of high stringency of the hybridization step for the purposes of defining the polynucleotide fragments described above are advantageously the following.

The DNA-DNA or DNA-RNA hybridization is carried out in two steps: (1) prehybric-lization at 42 C for 3 hours in phosphate buffer (20 mM, pH 7.5) containing 5 x SSC (1 x SSC corresponds to a 0.15 M NaC1 + 0.015 M sodium citrate solution), 50 % of formamide, 7 % of sodium dodecyl sulfate (SDS), 10 x Denhardt's, 5 %
of dextran sulfate and 1 % of salmon sperm DNA; (2) actual hybridization for 20 hours at a temperature dependent on the size of the probe (i.e.: 42 C, for a probe size > 100 nucleotides) followed by 2 washes of 20 minutes at 20 C in 2 x SSC + 2% of SDS, 1 wash of 20 minutes at 20 C in 0.1 x SSC + 0.1 % of SDS. The last wash is carried out in 0.1 x SSC + 0.1 % of SDS for 30 minutes at 60 C for a .probe size > 100 nucleotides.
The hybridization conditions of high stringency described above for a polynucleotide of defined size can be adapted by the person skilled in the art for oligonucicotides of greater or smaller size, according to the teaching of Sambrook et al. (1989, Molecular cloning: a laboratory manual. 2nd Ed. Cold Spring Harbor).
The invention likewise relates to a vector comprising a nucleic acid according to the present invention.
The invention aims especially at cloning and/or expression vectors which contain a nucleotide sequence according to the invention.
The vectors according to the invention preferably contain elements which allow the expression and/or the secretion of the nucleotide sequences in a determined host 2 0 cell. The vector must therefore contain a promoter, signals of initiation and termination of translation, as well as appropriate regions of regulation of transcription.
It must be able to be maintained in a stable manner in the host cell and can optionally have particular signals which specify the secretion of the translated protein.
These different elements are chosen and optimized by the person skilled in the art as a function of the host cell used. To this effect, the nucleotide sequences according to the invention can be inserted into autonomous replication vectors in the chosen host, or be integrative vectors of the chosen host.
Such vectors are prepared by methods currently used by the person skilled in the art, and the resulting clones can be introduced into an appropriate host by standard methods, such as lipofection, electroporation, thermal shock, or chemical methods.

The vectors according to the invention arc, for example, vectors of plasmidie or viral origin. They are useful for transforming host cells in order to clone or to express the nucleotide sequences according to the invention.
The invention likewise comprises the host cells transformed by or comprising a vector according to the invention.
The host cell can be chosen from prokaryotic or eukaryotic systems, for example bacterial cells but likewise yeast cells or animal cells, in particular mammalian cells. It is likewise possible to use insect cells or plant cells.
The invention likewise concerns animals, except man, which comprise at least one cell transformed according to the invention.
According to another aspect, a subject of the invention is a process for production of an antibody, or one of its functional fragments according to the invention, characterized in that it comprises the following stages:
a) culture in a medium and appropriate culture conditions of a host cell according to the invention; and b) the recovery of said antibodies, or one of their functional fragments, thus produced starting from the culture medium or said cultured cells.
The cells transformed according to the invention can be used in processes for preparation of recombinant polypeptides according to the invention. The processes for preparation of a polypeptide according to the invention in recombinant form, characterized in that they employ a vector and/or a cell transformed by a vector according to the invention, are themselves comprised in the present invention.

Preferably, a cell transformed by a vector according to the invention is cultured under conditions which allow the expression of said polypeptide and said recombinant peptide is recovered.
As has been said, the host cell can be chosen from prokaryotic or eukaryotic systems. In particular, it is possible to identify nucleotide sequences according to the invention, facilitating secretion in such a prokaryotic or eukaryotic system.
A vector according to the invention carrying such a sequence can therefore advantageously be used for the production of recombinant proteins, intended to be secreted. In effect, the purification of these recombinant proteins of interest will be facilitated by the fact that they are present in the supernatant of the cell culture rather than in the interior of the host cells.
It is likewise possible to prepare the polypeptides according to the invention by chemical synthesis. Such a preparation process is likewise a subject of the invention.
The person skilled in the art knows the processes of chemical synthesis, for example the techniques employing solid phases [Steward ct at., 1984, Solid phase peptide synthesis, Pierce Chem. Company, Rockford, 111, 2nd ed., (1984)] or techniques using partial solid phases, by condensation of fragments or by a classical synthesis in solution. The polypeptides obtained by chemical synthesis and being able to contain corresponding unnatural amino acids are likewise comprised in the invention.
The antibodies, or one of their functional fragments or derivatives, capable of being obtained by a process according to the invention arc likewise comprised in the present invention.
The invention also concerns the antibody of the invention as a medicament.
The invention likewise concerns a pharmaceutical composition comprising by way of active principle a compound consisting of an antibody, or one of its functional fragments according to the invention, preferably mixed with an excipient and/or a pharmaceutically acceptable vehicle.
Another complementary embodiment of the invention consists in a composition such as described above which comprises, moreover, as a combination product for simultaneous, separate or sequential use, an anti-tumoral antibody.
Most preferably, said second anti-tumoral antibody could be chosen through anti-IGF-1R, anti-EGFR, anti-IIER2/neu, anti-VEGFR, anti-VEGF, etc., antibodies or any other anti-tumoral antibodies known by the man skilled in the art. It is evident that the use, as second antibody, of functional fragments or derivatives of above mentioned antibodies is part of the invention.
As a most preferred antibody, anti-EGFR antibodies arc selected such as for Thl example the antibody C225 (Erbitux).
"Simultaneous use" is understood as meaning the administration of the two compounds of the composition according to the invention in a single and identical pharmaceutical form.
"Separate use" is understood as meaning the administration, at the same time, of the two compounds of the composition according to the invention in distinct pharmaceutical forms.
"Sequential use" is understood as meaning the successive administration of the two compounds of the composition according to the invention, each in a distinct pharmaceutical form.
In a general fashion, the composition according to the invention considerably increases the efficacy of the treatment of cancer. In other words, the therapeutic effect of the anti-c-Met antibodies according to the invention is potentiated in an unexpected manner by the administration of a cytotoxic agent. Another major subsequent advantage produced by a composition according to the invention concerns the possibility of using lower efficacious doses of active principle, which allows the risks of appearance of secondary effects to be avoided or to be reduced, in particular the effects of the cytotoxic agent.
In addition, this composition according to the invention would allow the expected therapeutic effect to be attained more rapidly.
The composition of the invention can also be characterized in that it comprises, moreover, as a combination product for simultaneous, separate or sequential use, a cytotoxic/cytostatic agent.
By "anti-cancer therapeutic agents" or "cytotoxic/cytostatic agents", it is 2 0 intended a substance which, when administered to a subject, treats or prevents the development of cancer in the subject's body. As non limitative example of such agents, it can be mentioned allcylating agents, anti-metabolites, anti-tumor antibiotics, mitotic inhibitors, chromatin function inhibitors, anti-angiogenesis agents, anti-estrogens, anti-androgens or immunomodulators.
Such agents are, for example, cited in the 2001 edition of VIDAL, on the page devoted to the compounds attached to the cancerology and hematology column "Cytotoxics", these cytotoxic compounds cited with reference to this document are cited here as preferred cytotoxic agents.
More particularly, the following agents are preferred according to the invention.
"Alkylating agent" refers to any substance which can cross-link or alkylate any molecule, preferably nucleic acid (e.g., DNA), within a cell. Examples of alkylating agents include nitrogen mustard such as mechlorethamine, chlorambucol, melphalen, chlorydrate, pipobromen, preclnimustin, disodic-phosphate or estramustine;
oxazophorins such as cyclophosphamide, altretamine, trofosfamide, sulfofosfamide or ifosfamide; aziridines or imine-ethylenes such as thiotepa, triethylenamine or altetramine; nitrosourea such as carmustine, streptozocin, fotemustin or lomustine;
5 alkyle-sulfonates such as busul fan, treosulfan or innprosulfan; triazen es such as dacarbazine; or platinum complexes such as cis-platinum, oxaliplatin and carboplatin.
"Anti-metabolites" refer to substances that block cell growth and/or metabolism by interfering with certain activities, usually DNA synthesis. Examples of anti-metabolites include methotrexate, 5-fluoruracil, floxuridine, 5-fluorodeoxyuridine, 10 capecitabine, cytarabine, fludarabine, cytosine arabinoside, 6-mercaptopurine (6-MP), 6-thioguanine (6-TG), chlorodesoxyadenosine, 5-azacytidine, gemcitabine, cladribine, deoxycoformycin and pentostatin.
"Anti-tumor antibiotics" refer to compounds which may prevent or inhibit DNA, RNA and/or protein synthesis. Examples of anti-tumor antibiotics include doxorubicin, 15 daunorubicin, idarubicin, valrubicin, mitoxantrone, dactinomycin, mithramycin, plicamycin, mitomycin C, bleomycin, and procarbazine.
"Mitotic inhibitors" prevent normal progression of the cell cycle and mitosis.
In general, microtubule inhibitors or taxoides such as paclitaxel and docetaxel are capable of inhibiting mitosis. Vinca alkaloid such as vinblastine, vincristine, vindesine and 20 vinorelbine are also capable of inhibiting mitosis.
"Chromatin function inhibitors" or "topoisomerase inhibitors" refer to substances which inhibit the normal function of chromatin modeling proteins such as topoisomerase I or topoisomerase II. Examples of chromatin function inhibitors include, for topoisomerasc I, camptothecine and its derivatives such as topotecan or irinotecan, 25 and, for topoisomerase II, etoposide, etoposide phosphate and teniposide.
"Anti-angiogenesis agent" refers to any drug, compound, substance or agent which inhibits growth of blood vessels. Exemplary anti-angiogenesis agents include, but are by no means limited to, razoxin, marimastat, batimastat, prinomastat, tanomastat, ilomastat, CGS-27023A, halofuginon, COL-3, neovastat, BMS-275291, thalidomide, 30 CDC 501, DMXAA, L-651582, squalamine, endostatin, SU5416, SU6668, interferon-alpha,EMD121974, interleukin-12, IM862, angiostatin and vitaxin.
"Anti-estrogen" or "anti-estrogenic agent" refer to any substance which reduces, antagonizes or inhibits the action of estrogen. Examples of anti-estrogen agents are tamoxifen, toremifene, raloxifene, droloxifene, iodoxyfene, anastrozole, letrozole, and cxemestane.
"Anti-androgens" or "anti-androgen agents" refer to any substance which reduces, antagonizes or inhibits the action of an androgen. Examples of anti-androgens are flutamide, nilutamide, bicalutamide, sprironolactone, cyproterone acetate, finastcride and cimitidinc.
"Immunomodulators" arc substances which stimulate the immune system.
Examples ofimmunomodulators include interferon, interleukin such as aldeslcukine, OCT-43, dcnileukin diflitox and interleukin-2, tumoral necrose fators such as tasoncrmine or others immunomodulators such as lentinan, sizofiran, roquinimex, pidotimod, pegademase, thymopentine, poly 1:C or levamisole in conjunction with 5-fluorouracil.
For more detail, the man skill in the art could refer to the manual edited by the "Association Francaisc des Enscignants de Chimic Therapcutique" and entitled "traite de chimie therapeutique, vol. 6, Medicaments antitumoraux et perspectives dans le traitcment des cancers, edition TEC & DOC, 2003".
Can also be mentioned as chemical agents or cytotoxic agents, all kinasc inhibitors such as, for example, gefitinib or erlotinib.
2 0 In a particularly preferred embodiment, said composition as a combination product according to the invention is characterized in that said cytotoxic agent is coupled chemically to said antibody for simultaneous use.
In order to facilitate the coupling between said cytotoxic agent and said antibody according to the invention, it is especially possible to introduce spacer molecules between the two compounds to be coupled, such as poly(alkylene) glycols like polyethylene glycol, or else amino acids, or, in another embodiment, to use active derivatives of said cytotoxic agents into which would have been introduced functions capable of reacting with said antibody according to the invention. These coupling techniques are well known to the person skilled in the art and will not be expanded upon in the present description.

The invention relates, in another aspect, to a composition characterized in that one, at least, of said antibodies, or one of their functional fragments or derivatives, is conjugated with a cell toxin and/or a radioelement.
Preferably, said toxin or said radioelemcnt is capable of inhibiting at least one cell activity of cells expressing the c-Met, in a more preferred manner capable of preventing the growth or the proliferation of said cell, especially of totally inactivating said cell.
Preferably also, said toxin is an enterobacterial toxin, especially Pseudomonas exotoxin A.
The radioelements (or radioisotopes) preferably conjugated to the antibodies employed for the therapy are radioisotopes which emit gamma rays and preferably iodine131, yttrium90, golc1199, palladiumloo, copper, bismuth217 and antimony211. The radioisotopes which emit beta and alpha rays can likewise be used for the therapy.
By toxin or radioelement conjugated to at least one antibody, or one of its functional fragments, according to the invention, it is intended to indicate any means allowing said toxin or said radioelement to bind to said at least one antibody, especially by covalent coupling between the two compounds, with or without introduction of a linking molecule.
Among the agents allowing binding in a chemical (covalent), electrostatic or 2 0 noncovalent manner of all or part of the components of the conjugate, mention may particularly be made of benzoquinone, carbodiimide and more particularly EDC
(1-ethy1-343-dimethyl-aminopropyli-carbodiimide hydrochloride), dimaleimidc, dithiobis-nitrobenzoic acid (DTNB), N-succinimidyl S-acetyl thio-acetate (SATA), the bridging agents having one or more phenylazide groups reacting with the ultraviolcts (U.V.) and preferably N4-4-(azidosalicylamino)buty1]-3'-(2'-pyridyldithio)-propionamide (APDP), N-succinimid-yl 3-(2-pyridyldithio)propionate (SPDP), 6-hydrazino-nicotinamide (HYN1C).
Another form of coupling, especially for the radioelements, can consist in the use of a bifunctional ion chelator.
Among these chelates, it is possible to mention the chelates derived from EDTA
(ethylenediaminetetraacetic acid) or from DTPA (diethylenetriaminepentaacetic acid) which have been developed for binding metals, especially radioactive metals, and immunoglobulins. Thus, DTPA and its derivatives can be substituted by different groups on the carbon chain in order to increase the stability and the rigidity of the ligand-metal complex (Krejcarek et al. (1977); Brechbiel et al. (1991); Gansow (1991);
US patent 4,831,175).
For example diethylenetriaminepentaacetic acid (DTPA) and its derivatives, which have been widely used in medicine and in biology for a long time either in their free form, or in the form of a complex with a metallic ion, have the remarkable characteristic of forming stable chclates with metallic ions and of being coupled with proteins of therapeutic or diagnostic interest such as antibodies for the development of radioimmunoconjugates in cancer therapy (Mcases et al., (1984); Gansow et al.
(1990)).
Likewise preferably, said at least one antibody forming said conjugate according to the invention is chosen from its functional fragments, especially the fragments amputated of their Fe component such as the scFv fragments.
As already mentioned, in a preferred embodiment of the invention, said cytotoxic/cytostatic agent or said toxin and/or a radioclement is coupled chemically to at least one of the elements of said composition for simultaneous use.
The present invention comprises the described composition as a medicament.
The present invention moreover comprises the use of the composition according to the invention for the preparation of a medicament.
In another aspect, the invention deals with the use of an antibody, or one of its functional fragments or derivatives, and/or of a composition as above described for the preparation of a medicament intended to inhibit the growth and/or the proliferation of tumor cells.
Another aspect of the invention consists in the use of an antibody, or one of its functional fragments or derivatives and/or of a composition, as described above or the use above mentioned, for the preparation of a medicament intended for the prevention or for the treatment of cancer.
Is also comprises in the present invention a method intended to inhibit the growth and/or the proliferation of tumor cells in a patient comprising the administration to a patient in need thereof of an antibody, or one of its functional fragments or derivatives according to the invention, an antibody produced by an hybridoma according to the invention or a composition according to the invention.

The present invention further comprises a method for the prevention or the treatment of cancer in a patient in need thereof, comprising the administration to the patient of an antibody, or one of its functional fragments or derivatives according to the invention, an antibody produced by an hybridoma according to the invention or a composition according to the invention.
In a particular preferred aspect, said cancer is a cancer chosen from prostate cancer, osteosarcornas, lung cancer, breast cancer, endometrial cancer, glioblastoma or colon cancer.
As explained before, an advantage of the invention is to allow the treatment of IMF dependent and independent Met-activation related cancers.
The invention, in yet another aspect, encompasses a method of in vitro diagnosis of illnesses induced by an overexpression or an undercxpression of the c-Met receptor starting from a biological sample in which the abnormal presence of c-Met receptor is suspected, said method being characterized in that it comprises a step wherein said biological sample is contacted with an antibody of the invention, it being possible for said antibody to be, if necessary, labeled.
Preferably, said illnesses connected with an abnormal presence of c-Met receptor in said diagnosis method will be cancers.
Said antibody, or one of its functional fragments, can be present in the form of an immunoconjugate or of a labeled antibody so as to obtain a detectable and/or quantifiable signal.
The antibodies labeled according to the invention or their functional fragments include, for example, antibodies called immunoconjugatcs, which can be conjugated, for example, with enzymes such as peroxidasc, alkaline phosphatase, beta-D-galactosidase, glucose oxydase, glucose amylase, carbonic anhydrase, acetylcholinesterase, lysozyme, malate dehydrogenase or glucose 6-phosphate dehydrogenasc or by a molecule such as biotin, digoxygenin or 5-bromodeoxyuridine. Fluorescent labels can be likewise conjugated to the antibodies or to their functional fragments according to the invention and especially include fluorescein and its derivatives, fluorochrome, rhodamine and its derivatives, GFP (GFP for "Green Fluorescent Protein"), dansyl, umbellifcrone etc. In such conjugates, the antibodies of the invention or their functional fragments can be prepared by methods known to the person skilled in the art. They can be coupled to the enzymes or to the fluorescent labels directly or by the intermediary of a spacer group or of a linking group such as a polyaldehyde, like glutaraldehyde, ethylenediaminetetraacetic acid (EDTA), diethylene-triaminepcntaacetic acid (DPTA), or in the presence of coupling agents such as those mentioned above for the therapeutic 5 conjugates.
The conjugates containing labels of fluorescein type can be prepared by reaction with an isothiocyanate.
Other conjugates can likewise include chemoluminescent labels such as luminol and the dioxetanes, bio-luminescent labels such as luciferase and luciferin, or else radioactive labels such as iodine123, iodine125, iodinci26, iodine133, bromine77, 10 technetium99m, indium''', j11j13m, gallium67, gallium68, ruthenium95, ruthenium97, ruthenium.m3, ruthenium"5, mercuryl 7, mercury203, rhenium89m, rhenium161, rhenium105, scandium47, tellurium121m, tellurium122m, tellurium125m, thulium165, thulium' 67, thIlliUMI68, fluorine", yttrium199, iodinc131. The methods known to the person skilled in the art existing for coupling the therapeutic radioisotopes to the antibodies either 15 directly or via a chelating agent such as EDTA, DTPA mentioned above can be used for the radioelcments which can be used in diagnosis. It is likewise possible to mention labeling with Na[I125] by the chloramine T method [Hunter W.M. and Greenwood F.C.
(1962) Nature 194:495] or else with technetium99m by the technique of Crockford et al.
(US patent 4,424,200) or attached via DTPA as described by Hnatowich (US
patent 20 4,479,930).
Thus, the antibodies, or their functional fragments, according to the invention can be employed in a process for the detection and/or the quantification of an overexpression or of an underexpression, preferably an overexpression, of the c-Met receptor in a biological sample, characterized in that it comprises the following steps:
25 a) the contacting of the biological sample with an antibody, or one of its functional fragments, according to the invention; and b) the demonstration of the c-Met/antibody complex possibly formed.
In a particular embodiment, the antibodies, or their functional fragments, according to the invention, can be employed in a process for the detection and/or the 30 quantification of the c-Met receptor in a biological sample, for the monitoring of the efficacy of a prophylactic and/or therapeutic treatment of c-Met-dependent cancer.

WO 2009/007427 PCT/EP2008/05902( More generally, the antibodies, or their functional fragments, according to the invention can be advantageously employed in any situation where the expression of the c-Met- receptor must be observed in a qualitative and/or quantitative manner.
Preferably, the biological sample is formed by a biological fluid, such as serum, whole blood, cells, a tissue sample or biopsies of human origin.
Any procedure or conventional test can be employed in order to carry out such a detection and/or dosage. Said test can be a competition or sandwich test, or any test known to the person skilled in the art depcndent on the formation of an immune complex of antibody-antigen type. Following the applications according to the invention, the antibody or one of its functional fragments can be immobilized or labeled. This immobilization can be carried out on numerous supports known to the person skilled in the art. These supports can especially include glass, polystyrene, poly-propylene, polyethylene, dcxtran, nylon, or natural or modified cells. These supports can be either soluble or insoluble.
By way of example, a preferred method brings into play immunocnzymatic processes according to the ELISA technique, by immunofluorescence, or radio-immunoassay (RIA) technique or equivalent.
Thus, the present invention likewise comprises the kits or sets necessary for carrying out a method of diagnosis of illnesses induced by an overexpression or an underexpression of the c-Met receptor or for carrying out a process for the detection and/or the quantification of an overexpression or of an underexpression of the c-Met receptor in a biological sample, preferably an overexpression of said receptor, characterized in that said kit or set comprises the following elements:
a) an antibody, or one of its functional fragments, according to the invention;
b) optionally, the reagents for the formation of the medium favorable to the immunological reaction;
c) optionally, the reagents allowing the demonstration of c-Met/antibody complexes produced by the immunological reaction.
A subject of the invention is likewise the use of an antibody or a composition according to the invention for the preparation of a medicament intended for the specific targeting of a biologically active compound to cells expressing or overexpressing the c-Met receptor.

It is intended here by biologically active compound to indicate any compound capable of modulating, especially of inhibiting, cell activity, in particular their growth, their proliferation, transcription or gene translation.
A subject of the invention is also an in vivo diagnostic reagent comprising an antibody according to the invention, or one of its functional fragments, preferably labeled, especially radiolabeled, and its use in medical imaging, in particular for the detection of cancer connected with the expression or the overexpression by a cell of the c-Met receptor.
The invention likewise relates to a composition as a combination product or to an anti-c-Met/toxin conjugate or radioelemcnt, according to the invention, as a medicament.
Preferably, said composition as a combination product or said conjugate according to the invention will be mixed with an excipicnt and/or a pharmaceutically acceptable vehicle.
In the present description, pharmaceutically acceptable vehicle is intended to indicate a compound or a combination of compounds entering into a pharmaceutical composition not provoking secondary reactions and which allows, for example, facilitation of the administration of the active compound(s), an increase in its lifespan and/or in its efficacy in the body, an increase in its solubility in solution or else an improvement in its conservation. These pharmaceutically acceptable vehicles are well known and will be adapted by the person skilled in the art as a function of the nature and of the mode of administration of the active compound(s) chosen.
Preferably, these compounds will be administered by the systemic route, in particular by the intravenous route, by the intramuscular, intradermal, intraperitoneal or subcutaneous route, or by the oral route. In a more preferred manner, the composition comprising the antibodies according to the invention will be administered several times, in a sequential manner.
Their modes of administration, dosages and optimum pharmaceutical forms can be determined according to the criteria generally taken into account in the establishment of a treatment adapted to a patient such as, for example, the age or the body weight of the patient, the seriousness of his/her general condition, the tolerance to the treatment and the secondary effects noted.

Other characteristics and advantages of the invention appear in the continuation of the description with the examples and the figures wherein:
Figure 1: Examples of FACS profiles of the selected anti-c-Met antibodies;
Figures 2 A and 2B: In vitro inhibition of BXPC3 proliferation by antibodies targeting e-Met;
Figure 3: Inhibition of c-Met dimerization;
Figure 4: Protein recognition by anti-c-Met antibodies;
Figures 5A and 5B: "Epitope mapping" of 11E1 and 5D5 by 13IAcore analysis;
Figures 6A and 6B: Effect of MAbs on c-Met phosphorylation;
Figures 7A and 7B: Displacement of radio-labeled HGF by anti-c-Met antibodies;
Figure 8: Inhibition of invasion by anti-c-Met antibodies [in this figure, SVF

means Fetal Calf Scrum (FCS)];
Figure 9: Effect of anti c-Met antibodies on wound healing;
Figures 10A and 10B: Scatter assay;
Figure 11: Three-dimensional tubulogenesis assay;
Figures 12A and 12B: Effect of antibodies on spheroid formation;
Figure 13: In vivo activity of anti-c-Met Mabs in the U87MG xenograft model;
Figure 14: HGF expression by a set of tumour cell lines;
Figures 15A and 15B: Characterization of the NCI-H441 cell line; with figure 15A corresponding to quantitative RT-PCR analysis and figure 15B corresponding to FACS analysis;
Figure 16: In vivo activity of anti-c-Met antibodies on NCI-H441 xenograft model;
Figure 17A: Alignment of 224G11 VL to murinc IGKV3-5*01 germlinc gene;
Figure 17B: Alignment of 224G1I VL to murine IGKJ4*0 I germline gene;
Figure 18A: Alignment of 224G11 VL to human 1GKV3-11*01 and 1GKV4-1*01 germline genes;
Figure 18B: Alignment of 224G1I VL to human IGKJ4*02 germline gene;
Figure 19A: IGKV3-11*01 based humanized version of 224G11 VL with mentioned mutations;
Figure 19B: IGKV4-1*01 based humanized version of 224G11 VL with mentioned mutations;
Figure 20A: Alignment of 224G11 VH to murine IGHV1-18*01 germline gene;
Figure 20B: Alignment of 224G11 VH to murine 1GHD2-4*01 gcnnline gene;
Figure 20C: Alignment of 224G11 VH to murine IGHJ2*01 germline gene;
Figure 21A: Alignment of 224G11 VH to human IGHV1-2*02 germlinc gene;
Figure 21B: Alignment of 224011 VH to human IGHJ4*01 germline gene;
Figure 22: Humanized 224011 VH with mentioned mutations;
Figure 23A: Alignment of 227H1 VL to murine IGKV3-5*01 germline gene;
Figure 23B: Alignment of 227H1 VL to murine IGKJ4*01 germline gene;
Figure 24A: Alignment of 227111 VL to human IGKV3-11*01 and IGKV4-1*01 germline genes;
Figure 24B: Alignment of 227H1 VL to human IGKI4*02 germline gene;
Figure 25A: IGKV3-11*01 based humanized version of 227H1 VL with mentioned mutations;
Figure 25B: IGKV4-1*01 based humanized version of 227H1 VL with mentioned mutations;
Figure 26A: Alignment of 227H1 VH to murine IGHVI-18*01 germline gene;
Figure 26B: Alignment of 227H1 VH to murine IGHD1-1*02 germline gene;
Figure 26C: Alignment of 227H1 VH to murine IGHJ2*01 germline gene;
Figure 27A: Alignment of 227H1 VH to human IGHV1-2*02 germline gene;
Figure 27B: Alignment of 227H1 VII to human IGHJ4*01 germline gene;
Figure 28: Humanized 227H1 VH with mentioned mutations;
Figure 29A: Alignment of 223C4 VL to murine TGKV12-46*01 germline gene;
Figure 29B: Alignment of 223C4 VL to murine IGKJ2*01 germline gene;
Figure 30A: Alignment of 223C4 VL to human IGKVI-NL1*01 germline gene;
Figure 30B: Alignement of 223C4 VL to human IGKJ2*01 germline gene:
Figure 31: Humanized 223C4 VL with mentioned mutations;
Figure 32A: Alignment of 223C4 VH to murine IGHV1-18*01 germline gene;
Figure 32B: Alignment of 223C4 VH to murine IGHD6-3*01 germline gene;
Figure 32C: Alignment of 223C4 VH to murine IGHJ4*01 germline gene;
Figure 33A: Alignment of 223C4 VH to human 1GHV1-2*02 germline gene;
Figure 33B: Alignment of 223C4 VH to human IGHD1-26*01 germlinc gene;

Figure 33C: Alignment of 223C4 VH to human IGHJ6*01 germline gene; and Figure 34: Humanized 223C4 VH with mentioned mutations;
Figure 35: Anti-tumor activity of the murine 224011 Mab alone or combined with Navelbine on the established xenograft NC1-H441 tumor model;
5 Figure 36: Evaluation of anti-c-Met Mabs on HUVEC
proliferation;
Figure 37: Evaluation of anti-c-Met Mabs on HUVEC tube formation;
Figure 38A: Alignment of 11E1 VL to murine IGKV4-79*01 germline gene;
Figure 38B: Alignment of 11E1 VL to murine IGKJ4*01 germline gene;
Figure 39A: Alignment of 11E1 VL to human IGKV3D-7*01 germline gene;
10 Figure 39B: Alignment of 11E1 VL to human IGKJ4*02 germline gene;
Figure 40: Humanized version of 11E1 VL with mentioned mutations;
Figure 41A: Alignment of 11E1 VH to murine IGHV1-7*01 germline gene;
Figure 41B: Alignment of 1 1E1 VH to =rine IGHD4-1*01 germline gene;
Figure 41C: Alignment of 11E1 VH to murine IGHJ3*01 germline gene;
15 Figure 42A: Alignment of 11E1 VH to human IGHV1-2*02 and IGHV1-46*01 germline genes;
Figure 42B: Alignment of 11E1 VH to human IGHJ4*03 germline gene;
Figure 43: Humanized 11E1 VH with mentioned mutations;
Figures 44A and 44B: c-Met Phosphorylation assay on A549 cells. Evaluation 20 of 11E1 and 224G11 purified Mabs, in absence or in presence of HGE, either at 30 lag/m1 (figure 44A) or within a dose range from 0.0015 to 30 ),mg/ml in order to determine EC50 values (figure 44B);
Figure 45: In vivo combination of 224011 Mab with Navelbine in the NSCLC
NCI-H441 xenograft model;
25 Figure 46: In vivo combination of 224011 Mab with Doxorubicin in the NSCLC
NCI-H441 xenograft model;
Figure 47: In vivo combination of 224G11 Mab with Docetaxcl in the NSCLC
NCI-H441 xcnograft model;
Figure 48: In vivo combination of 224G11 Mab with Temozolomide in the 30 NSCLC NCI-H441 xenograft model;
Figures 49A, 49B, 49C and 49D: Effect of anti-c-Met Mabs on U87-MG
spheroid growth;

Figures 50A and 50B: In vitro activity of chimeric and humanized forms of 224G11 in the phospho-c-Met assay;
Figure 51: Settings of Biacore analysis;
Figure 52: In vivo activity of 224G11 on MDA-MB-231 cells co-implanted with MRCS cells as human HGF source on Athymic nude mice;
Figure 53: ELISA based binding assay to Fc-cMct. Anti-Fe-c-Met binding activity was measured in an ELISA-based assay where anti-murinc Fe conjugates was used to detect the purified murinc monoclonal antibodies 11E1, 224G11 and 227H1.
Dose-dependent binding activities onto plastic-coated recombinant Fc-c1VIet was measured at 450nm;
Figure 54: HGF-cMet competition assay. In this ELISA-based assay, recombinant Fe-cMET residual binding to plastic coated HGF in the presence of purified murine monoclonal antibodies 11E1, 224G11 and 227H1 was detected with anti-murine Fc conjugate and measured at 450 nm;
Figure 55: Amino acid sequences alignment of 227H1-derived recombinant VH
domains. The 227H1 VH amino acid sequence is aligned with the selected human receiving framework sequence, with only mentioned the amino acids that were found different from the murine 227H1 VH sequence. 227H1 HZ1, HZ2 and HZ3 VH
sequences correspond to implemented humanized versions of the 227H1 murinc VH
domain, with remaining murinc residues in bold. In HZ3. 10 residues (*) were automatically changed for their human counterparts. In HZ2, the seven residues from the third group (3) have been studied. In HZ1VH, the nine residues from the second group (2) have been mutated into their human counterparts, only the six residues from the first group (1) remain murinc;
Figure 56: ELISA based binding assay to Fe-cMet of recombinant 227H1 antibodies. Anti-Fc-eMet binding activity was measured in an ELISA-based assay where anti-human Fe conjugates was used to detect chimeric and humanized 227H1-derived recombinant antibodies. Dose-dependent binding activities onto plastic-coated recombinant Fe-cMet of humanized VH domains-derived 227H1 antibodies was measured at 450nm and then compare to those of the parental/reference chimeric antibody;
Figure 57: ELISA based binding assay to Fc-cIVIet of recombinant 227H1-derived antibodies. Anti-Fc-cMet binding activity was measured in an ELISA-based assay where anti-human Fc conjugates was used to detect chimeric and humanized 227H1-derived recombinant antibodies. Dose-dependent binding activity onto plastic-coated recombinant Fc-cMet of humanized HZ4VH-derivcd 227H1 antibody was measured at 450nm and then compared to those of the parental/reference chimeric antibody;
Figure 58: HGF-cMet competition assay of 227111 murine and recombinant antibodies. In this ELISA-based assay, recombinant Fc-cMet residual binding to plastic coated HGF in the presence of the different forms of the 2271-11 antibody was detected with a biotinylated unrelated anti-cMct antibody. Purified murine 227H1 monoclonal antibody, chimeric and HZ4VII-derived humanized 227111-derived recombinant antibodies were tested and compared for their abilities to compete with HGF-cMet binding when measured at 450nm;
Figure 59: 227H1-HZ VH humanized variable domain sequence. *, corresponds to amino acids changed de facto to their human counterparts; 1, corresponds to amino acids humanized during the HZ3 to HZ! implementation; , corresponds to amino acids humanized in the final 227H1-HZ VH sequence;
Figure 60: Amino acid sequences alignment of 11E1-derived recombinant VH
domains. The 11E1 VH amino acid sequence is aligned with the selected human 2 0 receiving framework sequence, with only mentioned the amino acids that were found different from the murine 11E1 VH sequence. 11E1 HZ VH1, VH2 and VH3 sequences correspond to implemented humanized versions of the 11E1 murine VH domain, with remaining murine residues in bold. In HZ VH3, seven residues (*) were automatically changed for their human counterparts. In HZ VH2, the seven residues from the third group (3) have been studied. In HZ VH1, the five residues from the second group (2) have been mutated into their human counterparts, only the five residues from the first group (1) remain murine;
Figure 61: ELISA based binding assay to Fc-cMet of recombinant 11E1 antibodies. Anti-Fc-cMet binding activity was measured in an ELISA-based assay where anti-human Fe conjugates was used to detect chimeric and humanized 11E1-derived recombinant antibodies. Dose-dependent binding activities onto plastic-coated recombinant Fc-cMet of humanized VH domains-derived 11E1 antibodies was measured at 450nm and then compare to those of the parental/reference chimeric antibody;
Figure 62: Amino acid sequences alignment of 11E1-derived recombinant VL
domains. The 11E1 VL amino acid sequence is aligned with the selected human receiving framework sequence, with only mentioned the amino acids that were found different from the murine 11E1 VL sequence. 11E1 HZ VL1, VL2 and VL3 sequences correspond to implemented humanized versions of the 11E1 murine VL domain, with remaining murine residues in bold. In HZ VL3, ten residues (*) were automatically changed for their human counterparts. In HZ VL2, the eight residues from the third group (3) have been studied. In HZ VL1, the eight residues from the second group (2) have been mutated into their human counterparts, only the four residues from the first group (1) remain murine;
Figure 63: EL1SA based binding assay to Fc-cMet of recombinant 11E1 antibodies. Anti-Fe-eMet binding activity was measured in an EL1SA-based assay where anti-human Fe conjugates was used to detect chimeric and humanized 11E1-derived recombinant antibodies. Dose-dependent binding activities onto plastic-coated recombinant Fc-cMet of humanized VL domains-derived 11E1 antibodies was measured at 450nm and then compare to those of the parental/reference chimeric antibody;
Figure 64: ELISA based binding assay to Fc-cMet of recombinant 11E1 antibodies. Anti-Fc-cMet binding activity was measured in an ELISA-based assay where anti-human Fe conjugates was used to detect chimeric and humanized 11E1-derived recombinant antibodies. Dose-dependent binding activities onto plastic-coated recombinant Fc-cMet of single or double humanized domains-derived 11E1 antibodies was measured at 450nm and then compared to those of the parental/reference chimeric antibody;
Figure 65: Amino acid sequences alignment of 224G11 VH domain sequence.
The 224G11 VH amino acid sequence is aligned with the 227H1 VH sequence (underlined are non homologous residues) and with the selected human receiving framework sequence, with only mentioned the amino acids that were found different from the murine 224G11 VH sequence. 224G11 HZ VHO sequence correspond to "227H1-based/full-IMGT" humanized version of the 224G11 murine VH domain. In this sequence no outside-1MGT-CDRs residues remain murine;
Figure 66: EL1SA based binding assay to Fc-cMct of recombinant 224G11 antibodies. Anti-Fc-cMet binding activity was measured in an ELISA-based assay where anti-human Fc conjugates was used to detect chimeric and HZVHO-derived humanized 224G11-derived recombinant antibodies. Dose-dependent binding activity onto plastic-coated recombinant Fc-cMet of the HZVHO "full-IMGT" humanized VH
domain-derived 224G11 antibody was measured at 450nm and then compared to those of the parental/reference chimeric antibody;
Figure 67: fIGF-cMet competition assay of 224G11 murine and recombinant antibodies. In this ELISA-based assay, recombinant Fc-cMet residual binding to plastic coated HGF in the presence of the different forms of the 224G1 I antibody was detected with a biotinylated unrelated anti-eMet antibody. Purified murinc 224G1 1 monoclonal antibody, chimeric and HZVHO-derived humanized 224G11-derived recombinant antibodies were tested and compared for their abilities to compete with HGF-cMet binding when measured at 450nm;
Figure 68: Amino acid sequences alignment of 224G11 VL domain sequences.
The 224G1 1 VL amino acid sequence is aligned with the two selected human receiving framework sequences, with only mentioned the amino acids that were found different from the murine 224G11 VL sequence. 224G11 HZ VL3 sequence correspond to "shorter-CDR1" humanized version of the 224G11 murine VH domain while HZ VL6 correspond to the "longer-CDR I" version, with the remaining murine residues in bold.
For both basic humanized versions, the remaining murine residues are ranked for further humanization process where * corresponds to amino acids humanized in the basic versions, and 3, 2 and 1 correspond to the residues groups for the design of the implemented humanized versions;
Figure 69: ELISA based binding assay to Fc-cMet of recombinant 224G11 antibodies. Anti-Fc-cMet binding activity was measured in an EL1SA-based assay where anti-human Fe conjugates was used to detect chimeric and humanized 22G11-derived recombinant antibodies. Dose-dependent binding activities onto plastic-coated recombinant Fc-cMet of humanized VL3 and VL6 domains-derived 224G11 antibodies was measured at 450nm and then compare to those of the parental/reference chimeric antibody;

Figure 70: ELISA based binding assay to Fc-cMet of recombinant 224G11 antibodies. Anti-Fc-cMet binding activity was measured in an ELISA-based assay where anti-human Fc conjugates was used to detect chimeric and humanized 224G1 derived recombinant antibodies. Dose-dependent binding activities onto plastic-coated 5 recombinant Fc-cMet of humanized VL domains-derived 224G11 antibodies was measured at 450nm and then compare to those of the parental/reference chimeric antibody;
Figure 71: HGF-cMet competition assay of 224G11 marine and recombinant antibodies. in this ELISA-based assay, recombinant Fc-cMet residual binding to plastic 10 coated IIGF in the presence of the different forms of the 224G11 antibody was detected with a biotinylated unrelated anti-cMct antibody. Purified murine 224G11 monoclonal antibody, chimeric and HZ VL4-derived humanized 224G11-derived recombinant antibodies were tested and compared for their abilities to compete with HGF-cMet binding when measured at 450nm;
15 Figure 72: Amino acid sequence of VL4 humanized 224G11 VL domain sequence. *, corresponds to amino acids changed de facto to their human counterparts in the basic HZ VL6 version; !, corresponds to amino acids humanized during the to HZ VL4 implementation; , corresponds to amino acids that remain murine in the 224G11-HZ VL4 sequence;
20 Figure 73: ELISA based binding assay to Fc-cMet of recombinant antibodies. Anti-Fc-cMet binding activity was measured in an ELISA-based assay where anti-human Fe conjugates was used to detect chimeric and humanized 22G11-derived recombinant antibodies. Dose-dependent binding activities onto plastic-coated recombinant Fc-cMet of single- or double-humanized domains-derived 224G11 25 antibodies was measured at 450nm and then compare to those of the parental/reference chimeric antibody;
Figure 74: HGF-cMet competition assay of 224G11 murine and recombinant antibodies. In this ELISA-based assay, recombinant Fc-cMet residual binding to plastic coated HGF in the presence of the different forms of the 224G11 antibody was detected 30 with a biotinylated unrelated anti-cMet antibody. Purified murine 224G11 monoclonal antibody, chimeric and fully humanized 224G11-derived recombinant antibodies were tested and compared for their abilities to compete with HGF-cMet binding when measured at 450nm;
Figure 75: ELISA based binding assay to Fc-cMet of recombinant 224011 antibodies. Anti-Fc-cMet binding activity was measured in an ELISA-based assay where anti-human Fe conjugates was used to detect chimeric and humanized 22011-derived recombinant antibodies. Dose-dependent binding activities onto plastic-coated recombinant Fc-cMet of single mutants of the VL4-derived fully humanized antibodies was measured at 450nm and then compare to those of the parental/reference chimeric antibody;
Figure 76: ELTSA based binding assay to Fc-cMet of recombinant 224011 antibodies. Anti-Fc-cMet binding activity was measured in an ELISA-based assay where anti-human Fe conjugates was used to detect chimeric and humanized 22G11-derived recombinant antibodies. Dose-dependent binding activities onto plastic-coated recombinant Fc-cMet of single and multiple mutants of the VL4-derivcd fully humanized 224011 antibodies was measured at 450nm and then compare to those of the parental/reference chimeric antibody; and Figure 77: HGF-eMet competition assay of 224011 murinc and recombinant antibodies. In this ELISA-bascd assay, recombinant Fc-cMet residual binding to plastic coated HGF in the presence of the different forms of the 224011 antibody was detected with a biotinylated unrelated anti-cMet antibody. Purified murine 22401 1 monoclonal antibody, chimeric and single or multiple mutants of the VL4-derived fully humanized 224011 recombinant antibodies were tested and compared for their abilities to compete with HGF-cMet binding when measured at 450nm.
Example 1: Generation of antibodies against c-Met To generate anti-e-Met antibodies 8 weeks old BALB/c mice were immunized either 3 to 5 times subcutaneously with a CHO transfected cell line that express c-Met on its plasma membrane (20x106 cells/dose/mouse) or 2 to 3 times with a c-Met extracellular domain fusion protein (10-15 ,ug/dose/mouse) (R&D Systems, Catalog #
358MT) or fragments of this recombinant protein mixed with complete Freund adjuvant for the first immunization and incomplete Freund adjuvant for the following ones.
Mixed protocols in which mice received both CHO-cMet cells and recombinant proteins were also performed. Three days before cell fusion, mice were boosted i.p. or i.v. with WO 2009/007427 PC TfEP2008/059026 the recombinant protein or fragments. Then spleens of mice were collected and fused to SP2/0-Ag14 myeloma cells (ATCC) and subjected to HAT selection. Four fusions were performed. In general, for the preparation of monoclonal antibodies or their functional fragments, especially of murine origin, it is possible to refer to techniques which are described in particular in the manual "Antibodies" (Harlow and Lane, Antibodies: A
Laboratory Manual, Cold Spring Harbor Laboratory, Cold Spring Harbor NY, pp.
726, 1988) or to the technique of preparation of hybridomas described by Kohler and Milstein (Nature, 256:495-497, 1975).
Obtained hybridomas were initially screened by ELTSA on the c-Met recombinant protein and then by FACS analysis on A549 NSCLC, BxPC3 pancreatic, and U87-MG glioblastoma cell lines (representative profiles were presented in Figure 1) to be sure that the produced antibodies will be able to also recognize the native receptor on tumor cells. Positive reactors on these 2 tests were amplified, cloned and a set of hybridomas was recovered, purified and screened for its ability to inhibit in vitro cell proliferation in the BxPC3 model.
For that purpose 50 000 BxPC3 cells were plated in 96 well plates in RPMT
medium, 2 mM L. Glutamine, without SVF. 24 hours after plating, antibodies to be tested were added at a final concentration ranging from 0.0097 to 40 ug/m1 60 min before addition of 100 ng/ml of hHGF. After 3 days, cells were pulsed with 0.5 uCi of [31-I]thymidine for 16 hours. The magnitude of [3H]thymidine incorporated into trichloroacetic acid-insoluble DNA was quantified by liquid scintillation counting.
Results were expressed as raw data to really evaluate the intrinsic agonistic effect of each Mab (Figures 2A and 2B).
Then antibodies inhibiting at least 50% cell proliferation were evaluated as supernatants by BRET analysis on c-Met transfected cells. For that purpose, CHO stable cell lines expressing C-Met-Rluc or C-Met-Rluc and C-Met-K1100A-YFP were generated. Cells were distributed in white 96 well microplates in DMEM-F12/FBS
5 %
culture medium one or two days before BRET experiments. Cells were first cultured at 37 C with CO2 5 % in order to allow cell attachment to the plate. Cells were then starved with 200 ul DMEM/well overnight. Immediately prior to the experiment, DMEM was removed and cells quickly washed with PBS. Cells were incubated in PBS
in the presence or absence of antibodies to be tested or reference compounds, 10 min at 37 C prior to the addition of coelenterazine with or without HGF in a final volume of 50 uI. After incubation for further 10 minutes at 37 C, light-emission acquisition at 485 tun and 530 nm was initiated using the Mithras luminometer (Berthold) (1s/wave length/well repeated 15 times).
BRET ratio has been defined previously [Angers et al., Proc. Natl. Acad. Sci.
USA, 2000, 97:3684-3689] as: [(emission at 530 mn)-(emission at 485 nm) X
Cf]/(emission at 485 nm), where Cf corresponds to (emission at 530 nm) /
(emission at 485 nm) for cells expressing Rluc fusion protein alone in the same experimental conditions. Simplifying this equation shows that BRET ratio corresponds to the ratio 530/485 nm obtained when the two partners were present, corrected by the ratio 530/485 nm obtained under the same experimental conditions, when only the partner fused to R. reniformis luciferase was present in the assay. For the sake of readability, results arc expressed in milliBRET units (mBU); mBU corresponds to the BRET
ratio multiplied by 1000.
After this second in vitro test, 4 antibodies i) without intrinsic activity as a whole molecule in the functional test of proliferation, ii) inhibiting significantly BxPC3 proliferation (Figures 2A and 2B) and iii) inhibiting c-Met dimerization (Figure 3) were selected. These 3 antibodies of IgG1 kappa isotype were described as 11E1, 224G11, 223C4 and 227H1. In the experiments, the 5D5 Mab, generated by Genentech, and available at the ATCC, was added as a control for the intrinsic agonistic activity.
Figures 2A and 2B demonstrates that 11E1, 224G11, 223C4 and 227H1 were without any agonist activity in contrast to 5D5 which induced a dose dependent stimulation of cell proliferation in absence of ligand. A significant inhibition of cell proliferation was observed with the 4 selected antibodies. 5D5 is without effect on HGF-induced cell proliferation in this test.
When evaluated for blockade of dimerization significant effects reaching 32, 55, 69 and 52 % inhibition of dimerization for 224G11, 223C4,11E1 and 227H1 respectively were observed. Compared to basal signals in the respective experiments, 5D5 antibody is without effect in this dimerization model.
Example 2: Protein recognition by anti-c-Met antibodies To characterize the pattern of recognition of the 3 selected antibodies, 3 ELISA

have been set up with the recombinant c-Met protein, its monomeric fragment (obtained by cleavage of the recombinant c-Met-Fe protein and the recombinant SEMA
domain.
Results presented in figure 4 demonstrated that the 4 antibodies recognized both dimeric and monomeric proteins. To perform these ELISA the human dimeric c-Met protein (R&D sytems, cat# 358MT) is coated at the concentration of 0.7 ,ttg/nrd in PBS
overnight at 4 C. After saturation of the plates (Costar #3690) with a 0.5 %
gelatin solution 2 hours at 37 C, hybridoma supernatants arc incubated 1 hour at 37 C.
Once rinsed with PBS, the anti-mouse HRP-antibody (Jackson ImmunoResearch, catalog #115-035-164) is added to each well at a 1/5000 dilution in ELISA buffer (0.1 %
gelatin/0.05 % Tween 20 in PBS) and the plates incubated for 1 hour at 37 C.
After 3 washes in PBS, the activity of the peroxydase is revealed by the addition of 50 !al of TMB substrate (Uptima). The reaction is left to occur for 5 min at room temperature.
The reaction is stopped by the addition of 50 Ill/well of a 1 M H2SO4 solution and read on a plate reader at 450 nm. The same kind of protocol was performed on monomeric c-Met and SEMA domain but in that cases proteins were coated at 5 and 3 t1g/m1 respectively.
The 5D5 Mab introduced as a positive control recognized as expected the SEMA protein. 224G11, 227H1 and 223C4 did not bind the SEMA domain. 11E1 is able to bind the SEMA.
To determine whether 11E1 and 5D5, both recognizing the SEMA domain compete for overlapping epitopes, BlAcore analysis were performed. BlAcore system based on the Surface Plasmon Resonance phenomenon deliver data by monitoring binding events in real-time. It is then useful to group antibodies in a so called "cpitope mapping" experiments. A couple of antibodies unable to bind at the same time on the antigen molecule are classified in the same group (identical or neighbouring binding sites). At the opposite when their respective binding sites are sufficiently distant to allow a simultaneous binding of both antibodies these later are classified into two different groups. In such experiments, the antigen is commonly used as the ligand (immobilized on the sensorchip) and the antibodies arc used without any labelling as analytes (solution phase).
All the experiments described have been done on a BlAcore X instrument (GE
Healthcare Europe GmbH). A CM5 sensorchip (BlAcore) activated by a mouse anti-Tag-6His Mab (R&D System ref. MAB050) has been prepared following the manufacturer instructions by using the amine coupling kit (BlAcore). The running buffer (HBS-EP) and regeneration buffer (Glyeine, HC1) are from BIAcore. A
recombinant soluble version of the human HGF receptor produced as a chimeric 5 molecule c-Met-Fc-Tag His was from R&D systems (ref. 358-MT-CF). The experiments were done at 25 C, at a flow rate of 30 al/min. A 10 ,itg/m1 solution of c-Met in running buffer was injected during one minute on the flowce112 (fc2) typically 270 RU of the soluble form of c-Met were captured. The flowcelll (fel) was used as a reference to check any non specific binding of the antibodies to the sensorchip matrix.
1 0 Sequential injections of antibodies to be tested were performed. An antibody was injected on both flowcells during 2 minutes. A second antibody (or the same) was then injected in the same conditions. If no significant binding was observed a third injection was done with another antibody. The sensorchip was then regenerated by a single 30 s injection of the regeneration buffer. Either antibodies and c-Met-Fe were 1 5 discarded at this stage.
Analysis of the results:
The ability of an antibody "A" to block the binding of an antibody "B" is calculated by the ratio BIA/C¨(R2A/B/R1B)x100: where R2A/B is the response 20 corresponding to the binding of the MAb "B" when it was injected after Mab "A" and RIB is the response corresponding to the binding of the 1V1Ab "B" when it was injected first. A BIA/C below 20 % means that A is able to block the binding of B so that A and B have neighbouring binding sites.
The epitope mapping has been performed with 2 Mabs, 11E1 and 5D5.
Table 3 2nd Ab (B) 1st Ab (A) 11E1 6.5% 84.2%
5D5 98.4% 11.0%

Visualisation of the binding on around 270RU of captured c-Met-Fc by the sequential 2 minutes injections of Mabs 5D5 (first), 5D5 (second) and 11E1 (third) at a concentration of 10 g/ml each demonstrated that 5D5 and 11E1 bind clearly to two distant sites (Figure 5A). This observation was confirmed by the reciprocal sequence of antibody (Figure 5B).
Table 3 summarized the calculation ratio obtained with the different sequences of these 2 antibodies. Black values (over 75 %) mean that Mab A does not block the binding of Mab B. Bold/italic values (below 20 %) mean that the binding sites of both antibody (A and B) are identical or sufficiently close to unable a simultaneous binding.
Example 3: Effect of Mabs on c-Met phosphorylation To determine the activity of anti-c-Met antibodies on c-Met phosphorylation a phospho c-Met EL1SA assay was set-up. Briefly 500 000 A549 cells were seeded in each well of 6-well plates in F121( medium + 10 % KS. 16 hours before RCP' addition (100 ng/m1), cells were starved and each antibody to be tested was added at a final concentration of 30 ug/m1 15 minutes before ligand stimulation. 15 minutes after HGF
addition, cold lysis buffer was added, cells were scraped and cell lysates collected and centrifuged at 13 000 rpm for 10 mm at 4 C. Supernatants were quantified with a BCA
kit (Pierce) and stored at ¨20 C. For ELASA assay, a goat anti-c-Met antibody (R&D
ref. AF276) was used as a capture antibody (coating overnight at 4 C) and after a saturation step (1 h at RT) with a TBS-BSA 5 % buffer, 25 ug of protein from the different cell lysates was added to each well of the 96-well plate. After a 90 minute-incubation time at RT, plates were washed four times and an anti-phospho-c-Met antibody (Rabbit anti-pY1230-1234-1235 c-Met) was added. After an additional 1 hour incubation time and 4 washes an anti-rabbit-HRP (Biosource) was added for 1 hour at RT and then Luminol substrate was added before evaluation the luminescence with a Mithras device. Results presented in Figure 6B demonstrated that 11E1, 224011, 223C4 and 227H1 inhibit c-Met phosphorylation by 68, 54, 80 and 65 %
respectively compared to the 5D5 Mab which displayed a weaker inhibition of c-Met phosphorylation (42 %). In this test, a weak basal effect (less to 20 %) was observed with the 4 candidate antibodies (Figure 6A). As described in the various examples presented in this patent, this weak basal effect has no consequences on the activity of antibodies in other in vitro and in vivo tests. The 5D5 used as a control displayed, in this test a significant basal effect.
Example 4: Displacement of radio-labelled HGF by anti-c-Met antibodies To determine whether the anti-c-Met antibodies were able to displace HGF, binding experiments were set up. Briefly, protein A FlashPlate 96-well microplates (Perkin Elmer) were saturated with 0.5 % gelatine in PBS (200 Owen, 2 h at room temperature) before adding recombinant c-Met-Fe (R&D Systems) as a coating protein.
Two thousand ul of a 1 ug/m1 c-Met-Fe solution in PBS were added to each well.
Plates were then incubated overnight at 4 C. Free residual Protein A sites were further saturated with a non relevant hIgG (0.5 1.ig/well in PBS) for 2 h at room temperature.
Plates were washed with PBS after each step.
For competition assays, binding of [1251]-HGF (specific activity ¨ 2,000 Ci/mmol) at 200 pM to immobilized c-Met was measured in the presence of varying concentrations of the anti-c-Met monoclonal antibodies 11E1, 224G11, 223C4, or HGF (R&D Systems) ranging from 0.1 pM to 1 uM in PBS pH 7.4. The plates were incubated at room temperature for 6 h, then counted on a Packard Top Count Microplate Scintillation Counter. Non specific binding was determined in the presence of 1 uM of HGF. The monoclonal antibody 9G4, which is not directed at c-Met but specifically recognizes an E. cob' protein, was used as mouse IgG1 isotype control.
Percent of total specific [125I]-HG1F binding was plotted as a function of ligand concentration on semilog graphs. Concentrations of the various inhibitors required to inhibit the radioligand binding by 50 % (IC50) were determined graphically from the sigmoid competition curves obtained (Figures 7A and 7B).
As expected, non radiolabeled HGF was able to fully displace C2511-HGF
binding to immobilized c-Met, whereas the control antibody 9G4 did not show any HGF blocking activity (Figures 7A and 7B). Monoclonal anti-c-Met antibodies 11E1, 224G11, 223C4 and 227H1 were able to inhibit [12511-HGF binding to immobilized c-Met, with IC50 values of 20 nM, 3 nM, 2.7 nM and 5.8 nM, respectively. The values determined for antibodies 224G11, 223C4 and 227H1 were comparable to the IC50 value determined for non radiolabeled HGF, which was comprised between 3 and 5 nM, whereas antibody 1 1E1 exhibited a higher 1050 value.

Example 5: Inhibition of invasion by anti-c-Met antibodies To evaluate the inhibiting effect of the anti-c-Met antibodies on the invasion process, A549 cells were plated in the upper chamber of BD BioCoat" MatrigclTM

invasion chambers (6.5 mm diameter wells with 8-um pre size polycarbonate membranes). A459 cells were starved 24 hours before performing the invasion assay.
Then 500 000 A549 cells were plated in chemotaxis buffer (DMEM medium, 0.1 %
BSA, 12 mM Hepes) in the upper well of each chamber, upon the Matrigcl coating either with or without the antibody to be tested (final Mab coneentration10 ug/m1).
After 1 hour incubation of the plates at 37 C with 5 % CO2, the lower chambers were filled with either growth medium containing 400 ng/ml of rhHGE or with growth medium alone. The chambers were incubated for 48 additional hours at 37 C with 5 %
CO2. At the end of this incubation time, cells that remained on upper surface of the filter were gently removed with a cotton swab, cells that migrated to the lower surface of the filter were lysed, stained with CyQuant GR dye buffer (Invitrogen) and counted using a fluorescence reader Berthold Mithras LB940. All conditions were tested as triplicates.
As expected HGF induced a significant invasion of tumor cells comparable to the one observed with 10 % FCS introduced as a positive control (Figure 8).
The murine IgG1 9G4 introduced as an isotype control is without significant effect on basal or HGF-induced invasion when compared to cells plated without IgG. No agonist effect was noticed with 11E1, 224G11, 223C4 and 227H1 when added alone and a significant and comparable inhibition of the HGF-induced invasion was observed with the 3 Mabs.
Example 6: Inhibition of wound healing by anti-c-Met antibodies HGF stimulates motility. To determine whether the anti-HGF antibodies were able to inhibit migration, NCI-H441 cells were grown to high density and a gap was introduced with a P200 pipette tip. Cells were then stimulate to migrate across the gap with HGF (100 ng/ml) in presence or in absence of 11E1. Wells with 11E1 alone were also evaluated. Each tested condition was evaluated as a sextuplicate and 3 independent experiments were performed. After an overnight incubation, cells were visualized with an Axio Vision Camera (objective x4).
HGF induced a significant migration resulting in a complete closure of the gad WO 2009/007=127 PC T/EP2008/059026 within one night (Figure 9). The 9G4 irrelevant IgG I used as an isotype control is without any effect on cell migration. As expected an agonist effect was observed with the 5D5 when added alone but a significant inhibition of cell migration is observed with this antibody in presence of HGF in the portion of the gap remained open. The Fab fragment of 5D5 is without any agonist effect when added alone. However no activity of this fragment was observed in presence of HGF. As observed with the isotype control 9G4, the MAU 11E1 had no agonist effect when added alone and behave as a full antagonist in presence of FIGF.
Example 7: Scatter assay SK-HEP-1 cells were seeded at low density (1.104 cells/well) in a 24-well plate in DMEM with 10 % FCS and grown for 24 hours before addition, at the same time, of HGF (100 ng/ml) and antibodies to be tested (10 uWm1). After 72 hours incubation, colonies were fixed and stained with 0.2 % crystal violet in methanol and assessed for scattering visually. Each tested condition was tested as a triplicate and 3 independent experiments were performed.
Addition of HGF to SK-HEP-1 cells induced a significant cell scattering (Figures 10A and 10B). The 9G4 antibody introduced as an isotype control is without effect neither alone or in presence of HGF. As expected the 5D5 antibody displayed a significant agonist effect alone and no inhibitory effect was observed when 5D5 was added with HGF (Figure 10A). No agonistic effect was observed neither with (Figure 10A) nor with 224G11 (Figure 10B) added alone. A very significant inhibitory effect of these antibodies was demonstrated in presence of HGF (Figures 10A
and 10B).
Example 8: Three-dimensional tubulogenesis assay SK-HEP-1 cells were seeded at 1.104 cells/well in a 24-well plate in DMEM
with 10 % FCS/Matrigel (50/50) and incubated for 30 min before addition, at the same time, of HGF (100 ng/ml) and antibodies to be tested (10 .ng/ml). After 7 days incubation, cells were assessed for tube formation visually. Each tested condition was tested as a triplicate and 3 independent experiments were performed.
Addition of HGF induced a significant SK-HEP-1 tube formation (Figure 11).
The antibody 9G4 introduced as an isotype control was without effect neither alone or in presence of HGF. As expected the 5D5 antibody displayed a significant agonist effect alone and no inhibitory effect was observed when 5D5 was added with HGF. No agonistic effect was observed with 11E1, 223C4 and 224G11 added alone and a full inhibitory effect was demonstrate with both 11E1 and 223C4 in presence of HGF.
A
5 partial but significant inhibition was observed with the 224G11Mab.
Example 9: Spheroid formation To evaluate the ability of anti-c-Met antibodies to inhibit in vitro tumor growth, in a model closer to an in vivo situation, U-87MG, human glioblastoma cells (ATCC
10 HTB-14) spheroids were generated. Cells grown as a monolayer were detached with trypsine-EDTA and resuspended into complete cell culture media (DMEM) supplemented with 10 % FBS. Spheroids were initiated by inoculating 625 cells into single wells of round bottom, 96 plates in DiV1EM-10 % ECS. To prohibit cell adhesion to a substratum, the plates were pre-coated with polyHEMA in 95 % ethanol and air 15 dried at room temperature. The plates were incubated under standard cell culture conditions at 37 C, 5 % CO2 in humidified incubators. Purified monoclonal antibodies (10 gg/m1) were added after 3 and 7 days of spheroid culture. HGF (400 ng/ml) was added once after 4 days of culture. Spheroids were kept in culture for at least 10 days.
Then, spheroid growth was monitored by measuring the area of spheroids using 20 automeasure module of axiovision software. Area was expressed in um2. 8-16 spheroids were evaluated for each condition.
Figures 12A and 12B showed that in presence of 10 % FCS no stimulation was observed when HGF was added to the complete medium. As expected the 9G4 isotype control is without effect on spheroid growth. 11E1 and 223C4 reduced significantly 25 spheroid growth both in presence and in absence of HGF. No effect was observed with the 5D5 Fab fragment.
Example 10: In vivo activity of anti-c-Met Mabs in the U87MG xenograft model 30 Six to eight weeks old athymic mice were housed in sterilized filter-topped cages, maintained in sterile conditions and manipulated according to French and European guidelines. U87-MG, a glioblastoma cell line, expressing c-Met and autocrine for the ligand HGF, was selected for in vivo evaluations. Mice were injected subcutaneously with 5x106 cells. Then, six days after cell implantation, tumors were measurable (approximately 100 mm3), animals were divided into groups of 6 mice with comparable tumor size and treated twice a week with 1 mg/dose of each antibody to be tested. The mice were followed for the observation of xenograft growth rate and body weight changes. Tumor volume was calculated by the formula: t (Pi)/6 X length X
width X height.
The results obtained were summarized in Figure 13 and demonstrated that all tested antibodies inhibit significantly in vivo growth of U87-MG cells. The use of a neutralizing anti-IGF-1R antibody (IgG1) in panel A demonstrates that the observed in vivo inhibition is specifically related to a HGF-cMet axis modulation.
Example II: In vivo activity of anti-c-Met Mabs in the NCI-H441 xenograft model NCI-H441 is derived from papillary lung adenocarcinoma, expresses high levels of c-Met, and demonstrates constitutive phosphorylation of c-Met RTK.
To determine whether this cell line expresses high levels of c-Met and is able to produce HGF, both quantitative RT-PCRs and FACS or ELTSA (Quantikine HGF, R&D systems) were performed. For quantitative RT-PCRs, total HGF or cMet transcript expression levels in cell lines were assessed by quantitative PCR
using standard TaqManTm technique. HGF or c-Met transcript levels were normalized to the housekeeping gene Ribosomal protein, large, PO (RPLO) and results were expressed as normalized expression values (2-ddCT method).
The primer/probe sets for RPLO were forward, 5'-gaaactetgcattctegcttcetg-3' (SEQ ID No. 47); reverse, 5'-aggactcgtttgtacccgttga-3' (SEQ ID No. 48); and probe, 5'-(FAM)-tgcagattggctacccaactgttgca-(TAMRA)-3' (SEQ ID No. 49). The primer/probe sets for HGF were forward, 5'-aacaatgcctctggttcc-3' (SEQ ID No. 50);
reverse, 5'-cttgtagetgegtcetttac-3' (SEQ ID No. 51); and probe, 5'-(FAM)-ecttcaatagcatgtcaagtggagtga-(TAMRA)-3' (SEQ ID No. 52). The primer/probe sets for cMet were forward, 5'-cattaaaggagaccteaccatagctaat-3' (SEQ ID No. 53);
reverse, 5'-cctgatcgagaaaccacaacct-3' (SEQ ID No. 54); and probe, 5'-(FAM)-catgaagcgaccetctgatgtccca-(TAMRA)-3' (SEQ TD No. 55). The thermocycling protocol consisted of melting at 50 C for 2 minutes and 95 C for 10 minutes, followed by 40 cycles at 95 C for 15 seconds and 62 C for 1 minute.
No mRNA for HGF was found in NCI-I-1441 (Figure 14) and HGF is not detectable by ELISA in NCI-H441 supernatants. In these experiments U87-MG, a glioblastorna cell line known as an autocrine cell line for HGF, was introduced as a positive control. The RT-PCR analysis showed a significant level of HGF mRNA
in U87-MG and 1.9 ng HGF/million cells was detected in the supernatant of U87-MG
cells. Both quantitative RT-PCRs and FACS analysis Figures 15A and 15B
demonstrated that as expected NCI-H441 cells significantly overexpressed c-Met and that this expression was dramatically higher than the one observed for U87-MG
cells. In this experiment the MCF-7 cell line was introduced as a negative control.
Taken together NCI-H441 appears as a non autocrine constitutively activated cell line able to grow independently of HGF ligand in which a ligand-independent dimerization of c-met occurred as a consequence of the ovcrexprcssion of the receptor.
The evaluation of anti-e-met antibodies on the in vivo activity of this non autocrine cell line could give some insights about their potency to impact on c-met dimerization.
Figure 16 demonstrates that 224G11, 11E1 and 227H1 inhibited significantly in vivo growth of NCI-H441 suggesting that in addition to ligand dependent inhibition, these antibodies able to inhibit dimerization arc also able to target a ligand-independent inhibition of c-met. As mentioned above in the specification, with that last property, 224G11, [1E1 and 227E-11 are shown to be different from the 5D5 one armed (0A-5D5) anti-c-Met antibody.
Example 12: Humanization process by CDR-grafting of the antibody 1 - Humanization of the light chain variable domain Comparison of the nucleotidic sequence of the 224G11 VL with murine germline genes As a preliminary step, the nucleotidic sequence of the 224G11 VL was compared to the murine germline genes sequences part of the IMGT database.

Murine IGKV3-5*01 and IGKJ4*01 germline genes with a sequence identity of 99.31 % for the V region and 94.28 % for the J region, respectively, have been identified. Regarding the obtained identity, it has been decided to directly use the 224G11VL sequences to look for human homologies.
These alignments are represented in Figures 17A for the V gene and 17B for the J gene.
Comparison of the nucleotidic sequence of the 224G11 VI, with human germline genes In order to identify the best human candidate for the CDR grafting, the human germline gene displaying the best identity with the 224G11VL has been searched. To this end, the nucleotidic sequence of 224G1IVL has been aligned with the human germline genes sequences part of the I MGT database. For optimization of the selection, alignments between the protcic sequences were made to search for better homologies.
These two complementary methods led to the identification of two possible receiving human V sequences for the murine 224G11 VL CDRs. Nucleotidic alignment gives the human 1GKV3-11*01 germline gene with a sequence identity of 75.99 %
whereas proteic alignment gives the human IGKV4-1*01 germline gene with a sequence identity of 67.30 %. It worthnoting that in both cases, the two closest germline genes and the analysed sequences show different CDR1 amino acid lengths (10 amino acids in 224G11 VL; 6 amino acids in IGKV3-11*01; 12 amino acids in IGKV4-1*0 1).
For the J region, the best homology score was first obtained with human the human IGKJ3*01 showing a sequence identity of 80 %. But a higher number of consecutive identical nucleotides and a better amino acid fitting has been found in the alignment with human IGKJ4*02 germline gene (sequence identity of 77.14 %).
Thus the IGKJ4*02 germline gene was selected as receiving human J region for the murine 11E1 VL CDRs.
Alignments are represented in Figures 18A for the V region and 18B for the J
region.
Humanized version of 224G11 VL
Given the possibility of two receiving human V regions for the murine 224G11 VL CDRs, two humanized versions of the 224011 VI, domain will be described.
The first corresponds to an initial trial for a human framework with a shorter CDR1 length (IGKV3-11*01), the second with a longer CDRI length (1GKV4-1*01).
a) IGKV3-11*01 based humanized version of 224011 VL
The following steps in the humanization process consist in linking the selected gemiline genes sequences IGKV3-11*01 and IGKJ4*02 and also the CDRs of the nlurine 224011 VI, to the frameworks of these gerniline genes sequences.
As depicted in Figure 19A, the bolded residues in the 224011 VL sequence correspond to the twenty-five amino acids that were found different between VL domain and the selected human frameworks (Human FR, i.e. IGKV3-1 1 and IGKJ4*02).
Regarding to several criteriae such as their known participation in VI UVI, interface, in antigen binding or in CDR structure, the amino acid class changes between murine and human residues, localization of the residue in the 3D structure of the variable domain, three out of the twenty-five different residues have been identified to be eventually mutated. These three most important defined residues and mutations into their human counterparts being murine M39 into human L. 1-140 into A and R84 into 0.
(SEQ ID NO:72). These ranked one residues are shown in Figure 19A as bolded residues in the 224G11 HZ1VL sequence where they remained murine.
Of course, the above mentioned residues to be tested are not limited but must be considered as preferential mutations.
With the help of a molecular model, other mutations could be identified. Can be mentioned the following ranked two residues, i.e. residues 15 (LIP). 49 (P/A), 67 (L/R), 68 (E/A), 93 (PIS) and 99 (WE) on which mutations could also be envisaged in another preferred embodiment.
Of course, the above mentioned residues to be eventually tested are not limited but must be considered as preferential mutations. In another preferred embodiment, all the sixteen others ranked three residues among the twenty-five different amino acids could be reconsidered.
All the above mentioned mutations will be tested individually or according various combinations.
Figure 19A represents the implemented IGKV3-11*01 based humanized 224G11 VL with above mentioned mutations clearly identified. The number under each proposed mutation corresponds to the rank at which said mutation will be done.
b) 1GKV4-1*01 based humanized version of 224G11 VL
The following steps in the humanization process consist in linking the selected germline genes sequences IGKV4-1*01 and IGKJ4*02 and also the CDRs of the murine 224G1 I VL to the frameworks of these germline genes sequences.
As depicted in Figure 19B, the bolded residues in the 224G 11 VI., sequence corresponds to the twenty-two amino acids that were found different between VL domain and the selected human frameworks (Human FR, i.e. IGKV4-1*01 and 1 0 IGKJ4*02).
Regarding to several criteriae such as their known participation in VII/VI, interface, in antigen binding or in CDR structure, the amino acid class changes between murine and human residues, localization of the residue in the 3D structure of the variable domain, four out of the twenty-two different residues have been identified to be 1 5 eventually mutated. These four most important defined residues and mutations into their human counterparts being murine L4 into human M. M39 into L. H40 into A and into G (SEQ ID NO:73). These ranked one residues are shown in Figure 19B as bolded residues in the 224G11 HZ2VL sequence where they remained murine.
Of course, the above mentioned residues to be tested are not limited but must be 20 considered as preferential mutations.
With the help of a molecular model, other mutations could be identified. Can be mentioned the following ranked two residues, i.e. residues 25 (A/S), 66 (NI), 67 (L/R).
and 93 (P/S) on which mutations could also be envisaged in another preferred embodiment.
25 Of course, the above mentioned residues to be eventually tested are not limited but must be considered as preferential mutations. In another preferred embodiment. all the fourteen others ranked three residues among the twenty-two different amino acids could be reconsidered.
All the above mentioned mutations will be tested individually or according 30 various combinations.
Figure 19B represents the implemented IGKV4-1*01 based humanized 224011 VL with above mentioned mutations clearly identified. The number under each =

proposed mutation corresponds to the rank at which said mutation will be done.
II - Humanization of the heavy chain variable domain Comparison of the nucleotidic sequence of the 224G11 VI-1 with marine germline genes As a preliminary step, the nucleotidic sequence of the 224G11 VH was compared to the murine germline genes sequences part of the IMGT database Murine IGHV1-18*01, 1GHD2-4*01 and 1GHJ2*01 germline genes with a sequence identity of 92.70 % for the V region, 75.00 % for the D region and 89.36 %
for the J region, respectively, have been identified. Regarding the obtained identity, it has been decided to directly use the 224G11 VH sequences to look for human homologies.
These alignments are represented in Figures 20A for the V gene, 20B for the D
1 3 gene and 20C for the J gene.
Comparison of the nucleotidic sequence of the 224G11 VH with human gertnline genes In order to identify the best human candidate for the CDR grafting, the human germline gene displaying the best identity with the 224G11 VH has been searched. To this end, the nucleotidic sequence of 224G11 VH has been aligned with the human germline genes sequences part of the IMGT database. For optimization of the selection, alignments between the proteic sequences were made to search for better homologies.
These two complementary methods led to the identification of the same receiving human IGHV1-2*02 V sequence for the murine 224GI1 VH CDRs with a sequence identity of 75.00 % at the nucleotidic level and 64.30 % at the proteic level.
It is worthnoting that the D region strictly belongs to the CDR3 region in the VH
domain. The humanization process is based on a CDR-grafting approach.
Analysis of the closest human D-genes is not usefull in this strategy.
Looking for homologies for the J region led to the identification of the human IGHJ4*04 germline gene with a sequence identity of 78.72 %.
Human IGHVI-2*02 V germlinc gene and human IGH.I4*01 J germline gene have thus been selected as receiving human sequences for the murine 224G11 VH

CDRs.
Alignments are represented in Figure 21A for the V regionand 21B for the .1 region.
Thanctnized version 4224011 VII
The following steps in the humanization process consist in linking the selected germline genes sequences IGHV1-2*02 and IGHJ4*01 and also the CDRs of the murine 224011 VH to the frameworks of these germline genes sequences.
As depicted in Figure 22, the bolded residues in the 224011 VU sequence correspond to the thirty amino acids that were found different between 224011 VU
domain and the selected human frameworks (Human FR, i.e. IG11V1-2*02 and 1GHJ4*01).
Regarding to several eriteriae such as their known participation in VHIV1_ interface, in antigen binding or in CDR structure, the amino acid class changes between murine and human residues, localization of the residue in the 3D structure of the 1 5 variable domain, four out of the thirty different residues have been identified to he eventually mutated. These four most important defined residues and mutations into their human counterparts being murine D51 into human U. 055 into W, V80 into R and into T. These ranked one residues are shown in Figure 22 as bolded residues in the 224G11 HZVH sequence where they remained murine (SEQ ID NO:74).
Of course, the above mentioned residues to be tested are not limited but must he considered as preferential mutations.
With the help of a molecular model, other mutations could be identified. Can be mentioned the following ranked two residues, i.e. residues 25 (T/A), 48 (E/Q), 49 (SO).
53 (PA), 76 (AIV), 78 (I,/M) and 90 (DIE) on which mutations could also be envisaged in another preferred embodiment.
Of course, the above mentioned residues to be eventually tested arc not limited but must be considered as preferential mutations. In another preferred embodiment, all the nineteen others ranked three residues amonp, the thirty different amino acids could be reconsidered.
All the above mentioned mutations will be tested individually or according various combinations.
Figure 22 represents the humanized 224011 VU with above mentioned mutations clearly identified. The number under each proposed mutation corresponds to the rank at which said mutation will be done.
Example 13: Humanization process by CDR-grafting of the antibody 227111 1- Humanization of the light chain variable domain Comparison of the nucleotidic sequence of-the 227111 VL with murine germline genes As a preliminary step, the nucicotidic sequence of the 227H1 VL was compared to the murine germline genes sequences part of the IMGT database Murine IGKV3-5*01 and IGKJ4*01 germline genes with a sequence identity of 96.90 % for the V region and 97.29 % for the J region, respectively, have been identified. Regarding the obtained identity, it has been decided to directly use the 227H1 VI, sequences to look for human homologies.
These alignments are represented in Figures 23A for the V gene and 238 for the J gene.
Comparison of the nucleotidic sequence of the 22 7111 n with human germline genes In order to identify the best human candidate for the CDR grafting, the human germline gene displaying the best identity with the 2271-11 VL has been searched. To this end, the nucleotidic sequence of 227HI VI, has been aligned with the human germline genes sequences part of the IMGT database. For optimization of the selection, alignments between the proteie sequences were made to search for better homologies.
These two complementary methods led to the identification of two possible receiving human V sequences for the murinc 22 7111 VL CDRs. Nucleotidic alignment gives the human IGKV3-11*01 germline gene with a sequence identity of 7491 %
whereas protcic alignment gives the human IGKV4-1*01 gerrnline gene with a sequence identity of 64.00 %. It worthnoting that in both cases, the two closest germline genes and the analysed sequences show different CDR1 amino acid lengths (10 amino acids in 227H1 VL; 6 amino acids in IGKV3-11*01; 12 amino acids in IGKV4-1*01).
For the J region, the best homology score was first obtained with human thc,' human IGKJ3*01 showing a sequence identity of 78.38%. But a higher number of consecutive identical nucleotides and a better amino acid fitting has been found in the alignment with human IGKJ4*02 germline gene (sequence identity of 75.68 %).
Thus the IGKJ4*02 germline gene was selected as receiving human 1 region for the murine 227H1 VL CDRs.
Alignments are represented in Figures 24A for the V region and 24B for the J
region.
Humanized version of 22-1G II EL
Given the possibility of two receiving human V regions for the murine 227111 VL CDRs, two humanized versions of the 227H1 VL domain will be described. The first corresponds to an initial trial for a human framework with a shorter CDR1 length (1GKV 3-1 I *0 I ), the second with a longer CDR I length (IGK V4- 1 *0 1 ).
a) K V3-1 1 *0 1 based humanized version of 2271-I L
The following steps in the humanization process consist in linking the selected germline genes sequences IGKV3-1 1*01 and IGKJ4*02 and also the CDRs of the murine 227H1 VL to the frameworks of these germline genes sequences.
As depicted in Figure 25A, the bolded residues in the 227111 V1_, sequence corresponds to the twenty-six amino acids that were found different between domain and the selected human frameworks (Human FR, i.e. IGKV3-1 1*0 1 and IGKJ4*02).
Regarding to several critcriac such as their known participation in V11/VL
interface, in antigen binding or in CDR structure, the amino acid class changes between murine and human residues, localization of the residue in the 31) structure of the variable domain, three out of the twenty-six different residues have been identified to be eventually mutated. These three most important defined residues and mutations into their human counterparts being murine 139 into human L. 1140 into A and R84 into G.
These ranked one residues are shown in Figure 25A as bolded residues in the HZ1 VI, sequence where they remained murine (SEQ ID NO:75).
Of course, the above mentioned residues to be tested are not limited but must be considered as preferential mutations.
With the help of a molecular model, other mutations could be identified. Can be mentioned the following ranked two residues, i.e. residues 15 (LP), 25 (WA), 49 (P/A).
67 (UR), 68 (E/A), 93 (P/S) and 99 (S/12) on which mutations could also be envisaged in another preferred embodiment.
Of course, the above mentioned residues to be eventually tested are not limited but must be considered as preferential mutations. In another preferred embodiment, all the sixteen others ranked three residues among the twenty-five different amino acids could be reconsidered.
All the above mentioned mutations will be tested individually or according various combinations.
Figure 25A represents the implemented IGKV3-11*01 based humanized 227H1 VL with above mentioned mutations clearly identified. The number under each proposed mutation corresponds to the rank at which said mutation will be done.
b) IGKV4-1*01 based humanized version of 227111 VL
The following steps in the humanization process consist in linking the selected germline genes sequences 1GKV4-1*01 and IGKJ4*02 and also the CDRs of the murine 227H1 VI. to the frameworks of these germline genes sequences.
2_5 As depicted in Figure 25B, the bolded residues in the 227H1 VI, sequence corresponds to the twenty-four amino acids that were found different between V1_, domain and the selected human frameworks (Human FR, i.e. 1GKV4-1*01 and IGKJ4*02).
Regarding to several criteriae such as their known participation in VH/VL
interface, in antigen binding or in CDR structure, the amino acid class changes between murine and human residues, localization of the residue in the 3D structure of the variable domain, four out of the twenty-four different residues have been identified to be eventually mutated. These four most important defined residues and mutations into their human counterparts being murine L4 into human M, 139 into L. 1-140 into A and R84 into (3. These ranked one residues are shown in Figure 251-3 as bolded residues in the 2271-11 HZ2VL sequence where they remained murine (SEQ II) NO:76).
Of course, the above mentioned residues to be tested are not limited but must be considered as preferential mutations.
With the help of a molecular model, other mutations could be identified. Can be mentioned the following ranked two residues, i.e. residues 25 (V/S), 66 (WO, 67 (L/R).
and 93 (PIS) on which mutations could also be envisaged in another preferred em bodimcnt.

Of course, the above mentioned residues to be eventually tested are not limited but must be considered as preferential mutations. In another preferred embodiment, all the sixteen others ranked three residues among the twenty-two different amino acids could be reconsidered.
All the above mentioned mutations will be tested individually or according various combinations.
Figure 25B represents the implemented IGKV4-1*01 based humanized 227H1 VL with above mentioned mutations clearly identified. The number under each proposed mutation corresponds to the rank at which said mutation will be done.
II - Humanization of the heavy chain variable domain Comparison of the nucleoticlic sequence of the 227111 VII with murine germline genes As a preliminary step, the nucleoticfic sequence of the 227H1 VH was compared to the murinc germline genes sequences part of the IMGT database .
Murine IGHV1-18*01, IGHD1-1*02 and IGHJ2*01 germline genes with a sequence identity of 92.70 % for the V region, 63.63 % for the D region and 91.48 %
for the J region, respectively, have been identified. Regarding the obtained identity, it has been decided to directly use the 227HI VH sequences to look for human homologies.
These alignments are represented in Figures 26A for the V gene, 26B for the D
gene and 26C for the I gene.
Comparison of the nucleotidic sequence of the 227111 VII with human germline genes In order to identify the best human candidate for the CDR grafting, the human germline gene displaying the best identity with the 224011 VH has been searched. To this end, the nucicotidic sequence of 227H1 VET has been aligned with the human germline genes sequences part of the IMGT database. The receiving human IGFIV1-2*02 V sequence for the murine 224G1 I CDRs with a sequence identity o172.92 %
was thus identified.
It is worthnoting that the D region strictly belongs to the CDR3 region in the VH
domain. The humanization process is based on a CDR-grafting approach.
Analysis the closest human D-genes is not useful' in this strategy.
Looking for homologies for the .1 region led to the identification of the human IG11.14*0 1 germline gene with a sequence identity of 78.72 %.
Human IGHV1-2*02 V germline gene arid human IGHJ4*0 1 3 germline gene have thus been selected as receiving human sequences for the murine 227H1 VH CDRs.
Alignments are represented in Figures 27A for the V region and 27B for the J
region.
For optimisation of the selection, the man skilled in the art could also make alignments between the proteic sequences in order to help him in the choice.
Humanized version of 227111 1/7-1 The following steps in the humanization process consist in linking the selected germline genes sequences 1GHV1-2*02 and IGH.14*0 1 and also the CDRs of the murine 227111 VII to the frameworks of these germlinc genes sequences.
As depicted in Figure 28, the bolded residues in the 227111 VH sequence correspond to the thirty-two amino acids that were found different between domain and the selected human frameworks (Human FR, i.e. 1GHV1-2*02 and IGIU4*01).
Regarding to several eriteriae such as their known participation in VH/VL
interface, in antigen binding or in CDR structure, the amino acid class changes between murine and human residues, localization of the residue in the 3D structure of the variable domain, six out of the thirty-two different residues have been identified to be eventually mutated. These six most important defined residues and mutations into their human counterparts being murine L39 into human M. N40 into 11, L55 into W.166 into N, V80 into R and K82 into 1. These ranked one residues are shown in Figure 28 as bolded residues in the 227H1 HZVH sequence where they remained murine (SFQ ID NO:77).
Of course, the above mentioned residues to be tested are not limited but must be considered as preferential mutations.
With the help of a molecular model, other mutations could be identified. Can be mentioned the following ranked two residues, i.e. residues 48 (K/Q), 49 (110).
53 (l/M), 76 (A/V) and 78 (L/M) on which mutations could also be envisaged in another preferred embodiment.
Of course, the above mentioned residues to be eventually tested are not limited but must be considered as preferential mutations. In another preferred embodiment, all the twenty-one others ranked three residues among the thirty different amino acids could be reconsidered.
All the above mentioned mutations will be tested individually or according various combinations.
Figure 28 represents the humanized 227H1 VH with above mentioned mutations clearly identified. The number under each proposed mutation corresponds to the rank at which said mutation will be done.
Example 14: Humanization process by CDR-grafting of the antibody 223C4 I - Humanization of the light chain variable domain Comparison of the nucleotidic sequence of the 223C4 VL with murine germline genes As a preliminary step, the nucleotidic sequence of the 223C4 VI. was compared to the murine germline genes sequences part of the IMGT database Murine IGKVI2-46*01 and IGKJ2*Olgermline genes with a sequence identity of 99.64 % for the V region and 94.59 % for the J region, respectively, have been identified. Regarding the obtained identity, it has been decided to directly use the 223C4 VL sequences to look for human homologies.
These alignments are represented in Figures 29A for the V gene and 29B for the J gene.
Comparison of the nucleotidic sequence of the 223C4 VL with human germline genes In order to identify the best human candidate for the CDR grafting, the human germline gene displaying the best identity with the 223C4 VL has been searched. To this end, the nucleotidic sequence of 223C4 VL has been aligned with the human germline genes sequences part of the IMGT database.
Human IGKV1-NL1*01 and IGK12*01 germline genes with a sequence identity of 78.49 % for the V region and 81.08 % for the J region, respectively, have been identified. The germline genes IGKVI-NLI*01 for the V region and IGKJ2*01 for the J region have thus been selected as receiving human sequences for the murine VL CDRs.
Alignments are represented in Figures 30A for the V region and 30B for the J
region.
For optimisation of the selection, the man skilled in the art could also make alignments between the proteic sequences in order to help him in the choice.
Humanized version of 223C4 VL
The following steps in the humanization process consist in linking the selected germline genes sequences IGKV1-NL1*01 and IGKJ2*01 and also the CDRs of the murine 223C4 VL to the frameworks of these germline genes sequences.
At this stage of the process, a molecular model of the 223C4 murine Fv domains could be developed and useful in the choice of the murine residues to be conserved due to their roles in the maintenance of the three-dimensional structure of the molecule or in the antigen binding site and function. More particularly, 9 residues to be eventually mutated have been identified.
In a first step, residues involved in the CDR anchors or structure will be tested.
Such residus are residu 66 (R/N) and residu 68 (EN).
In a second step, residus exposed to solvant, and as such that may involve immunogenicity, will also be tested. These arc residues 49 (A'S), 51 (K/Q), 69 (S/D), 86 (D/Q) and 92 (S/N).
Then, in a third step, residus involved in structure/folding of variable domain could also be mutated. These residues are residu 46 (P/Q) and residu 96 (P/S).
Of course, the above mentioned residus to be tested are not limitated but must be considered as preferential mutations.
With the help of a molecular model, other mutations could be identified. Can be mcntionned the following residues, i.e. residues 9 (S,/A), 13 (A/V), 17 (DIE), 18 (R/T), 54 (LN), 88 (T/S), 90 (T/K), 100 ()VG) and 101 (T/S), on which mutations could also be envisaged in another preferred embodiment.
All the above mentioned mutations will be tested individually or according various combinations.
Figure 31 represents the humanized 223C4 VL with above mentioned mutations clearly identified. The number under each proposed mutation corresponds to the rank at which said mutation will be done.

II - Humanization of the heavy chain variable domain Comparison of the nucleoticlic sequence of the 223C4 VH with marine gel-Mine genes As a preliminary step, the nucleotidie sequence of the 223C4 VII was compared to the murine genntine genes sequences part of the IMGT database Murine IGIIV1-18*01, IGHD6-3*01 and IGHJ4*01 germline genes with a sequence identity of 98.95 % for the V region, 72.72 % for the D region and 98.11 %
for the J region, respectively, have been identified. Regarding the obtained identity, it 1 0 has been decided to directly use the 223C4 VH sequences to look for human homologies.
These alignments are represented in Figures 32A for the V gene, 32B for the D
gene and 32C for the J gene.
Comparison of the nucleoticlic sequence of the 223C4 VH with human germ/inc genes In order to identify the best human candidate for the CDR grafting, the human germline gene displaying the best identity with the 223C4 VH has been searched. To this end, the nucleotidic sequence of 223C4 VH has been aligned with the human germline genes sequences part of the IMGT database.
Human IGHV1-2*02, IGHD1-26*01 and IGHJ6*01 germline genes with a sequence identity of 76.38 % for the V region, 75.00 % for the D region and 77.41 %
for the J region, respectively, have been identified. The germline genes IGHV1-2*02 for the V region and IGHJ6*01 for the J region have thus been selected as receiving human sequences for the murine 223C4 VH CDRs.
Alignments are represented in Figures 33A for the V region, 33B for the D
region and 33C for the J region.
For optimisation of the selection, the man skilled in the art could also make alignments between the proteie sequences in order to help him in the choice.
Humanized version of 223C4 VII
The following steps in the humanization process consist in linking the selected gertnline genes sequences IGHV1-2*02 and IGHJ6*01 and also the CDRs of the murine 223C4 VH to the frameworks of these germline genes sequences.

At this stage of the process, a molecular model of the 223C4 murine Ev domains could be developed and useful in the choice of the murine residues to be conserved due to their roles in the maintenance of the three-dimensional structure of the molecule or in the antigen binding site and function. More particularly, 14 residues to be eventually mutated have been identified.
In a first step, residues involved in the CDR anchors or structure will be tested.
Such residues are residues 40 (fl/D), 45 (A/S), 55 (W/D), 66 (N/I) and 67 (Y/F).
In a second step, residues exposed to solvant, and as such that may involve immunogenicity, will also be tested. These are residues 1 (Q/E), 3 (Q/L), 5 (V/Q), 48 (Q/M) and 80 (R/V).
Then, in a third step, residues involved in structure/folding of variable domain could also be mutated. These are residues 9 (A/P), 13 (K/V), 22 (S/P) and 46 (P/I-I).
Of course, the above mentioned residus to be tested are not limitate.d but must be considered as preferential mutations.
1 5 With the help of a molecular model, other mutations could be identified. Can be mentionned the following residues, i.e. residues 12 (VI), 21 (V/I), 43 (R/K), 49 (G/S), 53 (M/I), 68 (A/N), 72 (Q/K), 75 (R/K), 76 (V/A), 78 (M/L), 82 (T/K), 84 (FS), (SIR), 93 (R/S), 95 (Rif) and 97 (DIE), on which mutations could also be envisaged in another preferred embodiment.
All the above mentioned mutations will be tested individually or according various combinations.
Figure 34 represents the humanized 223C4 VII with above mentioned mutations clearly identified (SEQ ID NO:78). The number under each proposed mutation corresponds to the rank at which said mutation will he done.
Example 15: Anti-tumor activity of the inurine 224C11 MAI) alone or combined with the chemotherapeutic agent Navelbinet on the established zenograft NC1-H441 tumor model Successful chemotherapeutic approaches depend in part on the cellular response to apoptotic inducers and the balance between pro- and anti-apoptotic pathways within the cell. The protective effect of the activated c-Met on cell survival has been documented. It mainly results from an increase expression of the anti-apoptotic Rcl-xl and Bc1-2 protein as a consequence of P13-K-mediated signaling which in turn inhibit mitochondrial-dependent apoptosis (caspase 9). Indeed, it is conceivable that the HGF/c-Met system with its marked regulatory effect on apoptotic process can also influence the chcmosensitivity of cancer cells. This hypothesis as been tested with Navelbine , a marketed chemotherapeutic agent used for lung cancer treatment (Aapro et al., Crit.Rev.Oncol.Hematol. 2001, 40:251-263; Curran et al., Drugs Aging.
2002, 19:695-697). The xcnograft NCI-1-1441 NSCLC model was used as it has been previously described that this cell line is sensitive to both Navelbine (Kraus-Berthier ct al., Clin.Cancer Res., 2000; 6:297-304) and therapy targeting c-Met (Zou H. T.
et al., Cancer Res. 2007, 67: 4408-4417).
Briefly, NCI-H441 cells from ATCC were routinely cultured in RPMI 1640 medium, 10 % FCS and 1 % L-Glutamine. Cells were split two days before engraftment so that they were in exponential phase of growth. Ten million NCI-H441 cells were engrafted in PBS to 7 weeks old Swiss nude mice. Three days after implantation, tumors were measured and animals were divided into 4 groups of 6 mice with comparable tumor size. Mice were treated i.p. with a loading dose of 2 mg of 224G11/mouse and then twice a week, for 43 days, with 1 mg of antibody/mouse.
The 9G4 MAb was used as an isotype control.
Navelbine was given by i.p. injections at a dose of 8 mg/kg on days 5, 12, 19 post-cell injection. For combined therapy with both 224G11 and Navelbine , the two compounds were administered separately. In this experience the 2 compounds were used at their optimal dosage. Tumor volume was measured twice a week and calculated by the formula: p/6 X length X width X height.
Figure 35 demonstrates that 224G11 is as efficient as Navelbine when used alone as a single agent therapy. A significant benefit of combining both therapy was observed with complete tumor regressions observed for 3 out of 6 mice at day 63.
Example 16: C-Met inhibitors and angiogenesis In addition to its direct role in the regulation of a variety of tumor cell functions, activation of c-met has also been implicated in tumor angiogenesis.
Endothelial cells expresse c-Met and HGF stimulates endothelial cell growth, invasion and motility (Nakamura Y. et al., Biochem. Biophys. Res., Commun. 1995, 215:483-488;
Bussolino F. et al., J. Cell Biol. 1992, 119:629-641). The coordinate regulation of growth, invasion and motility in vascular endothelial cells by HGF/c-Met has been demonstrated to results in the formation of 3D capillary endothelial tubes in vitro (Rosen E.M. et al., Supplementum to Experientia 1991, 59:76-88).
To determine a potential interference of anti-c-Met MAbs with HGF-induced angiogenesis, two sets of experiments were performed including i) the evaluation of MAbs on HUVEC proliferation and ii) the test of MAbs of HUVEC tube formation.
For proliferation experiments, 7500 HUVEC were plated in each well of a 96 well plate previously coated with laminin. Cells were grown 24 hours of EMB-2 assay medium supplemented with 0.5 % FBS and heparin. Then, MAbs to be tested (0.15 to 40 lag/m1) were added for 1 h before addition of 20 ng/ml of HUE. After 24 additional hours, cells were pulsed with 0.5 laCi of [3E1] Thymidine. The magnitude of CH]
Thymidine incorporated was quantified by liquid scintillation counting. In This experiment the 9G4 MAb is an irrelevant antibody used as an IgG1 isotype control.
Results expressed as raw data in figure 36 demonstrate that, as expected HGF
is a potent inducer of HUVEC cell growth. Antibodies evaluated in absence of HGF
did not display any agonist proliferative activity on HUVEC whatever the tested dose. In presence of HGF, a dramatic dose dependent inhibition was observed for both 11E1 and 224G11 MAbs For evaluation of HUVEC tube formation, 25000 cells incubated 30 min with antibodies to be tested were plated in 48-well plates coated with matrigel.
Then HGF 50 ng/ml was added and plates were incubated at 37 C. Medium was then harvested and 5 iM CMFDA was added for 15 min before microscopic observation.
Results shown in figure 37 demonstrate that, as expected HGF induces a significant tube formation. The 9G4 antibody introduced as an IgG1 isotype control was without any effect on HGF-induced tube formation whereas both 11E1 and 224G11 inhibit dramatically tube formation.
Example 17: Humanization process by CDR-grafting of the antibody 11E1 I - Humanization of the light chain variable domain Comparison of the nucleotidic sequence of the 11E1 VL with murine germline genes As a preliminary step, the nucleotidic sequence of the 11E1 VL was compared to the murine germline genes sequences part of the I7VIGT database .
Murine IGKV4-79*01 and IGKJ4*01 germline genes with a sequence identity of 98.58 % for the V region and 97.22 % for the J region, respectively, have been identified. Regarding the obtained identity, it has been decided to directly use the 11E1 VL sequences to look for human homologies.
These alignments arc represented in Figures 38A for the V gene and 38B for the J gene.
Comparison of the nuc1eotidic sequence of the 11E1 VI, with human germline genes In order to identify the best human candidate for the CDR grafting, the human germlinc gene displaying the best identity with the 11E1 VI, has been searched. To this end, the nucleotidic sequence of 11E1 VL has been aligned with the human germline genes sequences part of the 'MGT database.
Human 1GKV3-7*02 and IGKV3D-7*01 with a sequence identity for both gcrmline genes of 69.86 % for the V region have been identified. IGKV3-7O2 human germline gene is known in the IMGT database as an "ORF" which mean that this sequence has been found in the human genome but may present some recombination problems leading to non functional IGKV3-7*02 derived natural antibodies. Thus the 1GKV3D-7*01 germline gene was selected as receiving human V region for the murine 11F,1 VI , CDR s.
For the J region, the best homology score was first obtained with human, the human 1GKJ3*01 showing a sequence identity of 78.38 %. But a higher number of consecutive identical nucleotides and a better amino acid fitting has been found in the alignment with human IGKJ4*02 germline gene (sequence identity of 75.68 %).
Thus the IGKJ4*02 germline gene was selected as receiving human J region for thc murine 11E1 VL CDRs.
Alignments are represented in Figures 39A for the V region and 39B for the J
region.
For optimisation of the selection, the man skilled in the art could also make alignments between the proteic sequences in order to help him in the choice.

Humanized version oh/El FL
The following steps in the humanization process consist in linking the selected gerrnline genes sequences 1GKV3D-7*01 and 1GKJ4*02 and also the CDRs of the marine 11E1 VL to the frameworks of these germline genes sequences.
As depicted in Figure 40, the bolded residues in the 11E1 VL sequence corresponds to the thirty amino acids that were found different between 11E1 VL
domain and the selected human frameworks (Human FR, i.e. 1GKV3D-7*01 and 1GKJ4*02).
Regarding to several criteriae such as their known participation in VEUVL
interface, in antigen binding or in CDR structure, the amino acid class changes between murine and human residues, localization of the residue in the 3D structure of the variable domain, four out of the thirty differents residues have been identified to be eventually mutated. These four most important defined residues and mutations into their human counterparts being murine L4 into human M, Y40 into S. Y87 into F and into P. These ranked one residues are shown in Figure 40 as bolded residues in the 11E1 HZVL sequence where they remained murine (SEQ ID NO:79).
Of course, the above mentioned residues to be tested are not limited but must be considered as preferential mutations.
With the help of a molecular model, other mutations could be identified. Can be mentioned the following ranked two residues, i.e. residues 24 (S/R), 53 (W/L), 66 orn.
67 (L/R), 86 (S/D), 95 (Q/E), 99 (A/F) or 121 (F/D) on which mutations could also be envisaged in another preferred embodiment.
Of course, the above mentioned residues to be eventually tested are not limited but must be considered as preferential mutations. In another preferred embodiment, all the eighteen others ranked three residues among the thirty different amino acids could be reconsidered.
All the above mentioned mutations will be tested individually or according various combinations.
Figure 40 represents the implemented humanized 11E1 VL with above mentioned mutations clearly identified. The number under each proposed mutation corresponds to the rank at which said mutation will be done.

II - Humanization of the heavy chain variable domain Comparison of the nucleotidic sequence of the 11E1 VII with murine germline genes As a preliminary step, the nuelcotidic sequence of the 11E1 VH was compared to the murine germline genes sequences part of the IMGT database.
Murinc IGHV1-7*01, IGHD4-1*01 and 10HJ3*01 germline genes with a sequence identity of 94.10 % for the V region, 66.67 % for the D region and 100 % for the J region, respectively, have been identified. Regarding the obtained identity, it has been decided to directly use the 11E1 VH sequences to look for human homologies.
These alignments are represented in Figures 4IA for the V gene, 41B for the D
gene and 4IC for the J gene.
Comparison of the nucleotidic sequence of the 11E1 VH with human germ/inc genes In order to identify the best human candidate for the CDR grafting, the human germline gene displaying the best identity with the 11E1 VH has been searched.
To this end, the nucleotidic sequence of 11E1 VH has been aligned with the human germline genes sequences part of the IMGT database. For optimization of the selection, alignments between thc proteic sequences were made to search for better homologies.
These two complementary methods led to the identification of two possible receiving human V sequences for the marine 11E1 VH CDRs. Nucleotidic alignment gives the human IGHV1-2*02 germline gene with a sequence identity of 75.69 %
whereas protcic alignment gives the human IGHV1-46401 germline gene with a sequence identity 01'71.10 %.
It is worthnoting that the D region strictly belongs to the CDR3 region in the VH
domain. The humanization process is based on a CDR-grafling approach.
Analysis of the closest human D-genes is not uscfull in this strategy.
Looking for homologies for the J region led to the identification of the human IGHJ4*03 germline gene with a sequence identity of 80.85 %.
Looking to the overall similarities and sequences alignments, human IGHV1-46*01 V germline gene and human IGHJ4*03 J germline gene have thus been selected as receiving human sequences for the murine 11E1 VH CDRs.

Alignments are represented in Figures 42A for the V region and 4213 for the J
region.
Humanized version of 11E1 17H
The following steps in the humanization process consist in linking the selected germline genes sequences IGHV1-46*01 and I0H.14*03 and also the CDRs of the murine 11E1 VH to the frameworks of these germline genes sequences.
As depicted in Figure 43, the bolded residues in the 11E1 VH sequence corresponds to the twenty-six amino acids that were found different between domain and the selected human frameworks (Human FR, i.e. IGHV1-46*01 and IGHJ4*03).
Regarding to several criteriae such as their known participation in VII/VI.
interface, in antigen binding or in CDR structure, the amino acid class changes between murine and human residues, localization of the residue in the 3D structure of the variable domain, five out of the twenty-six differents residues have been identified to be 1 5 eventually mutated. These five most important defined residues and mutations into their human counterparts being murine N40 into human 14, Y55 into 1, D66 into S. A80 into R and K82 into T. These ranked one residues are shown in Figure 43 as bolded residues in the 11E1 FIZVH sequence where they remained murine (SEQ Ill NO:80).
Of course, the above mentioned residues to be tested are not limited but must be 2 0 considered as preferential mutations.
With the help of a molecular model, other mutations could be identified. Can be mentioned the following ranked two residues, i.e. residues 53 (I/M). 71 (LIE), 76 (A/V), 78 (I./M) and 87 (A/V) on which mutations could also be envisaged in another preferred embodiment.
25 Of course, the above mentioned residues to be eventually tested are not limited but must be considered as preferential mutations. In another preferred embodiment, all the sixteen others ranked three residues among the twenty-six different amino acids could be reconsidered.
All the above mentioned mutations will be tested individually or according 30 various combinations.
Figure 43 represents the implemented humanized 11E1 VH with above mentioned mutations clearly identified. The number under each proposed mutation WO 2009/007,127 PCT/EP2008/059026 corresponds to the rank at which said mutation will be done.
Example 18: Effect of purified Mabs on c-met phosphorylation In example 3, the effect of anti-c-Met Mabs on phosphorylation was assessed with dosed supernatants from each hybricloma to be evaluated. The test has been performed again with purified 11E1 and 224G11 Mabs that have been evaluated either at a final concentration of 30 tg/m1 (200 nM) or at a dose range from 0.0015 to 30 ng/ml (0.01-200 nM) in order to determine the IC50 of each antibody. The protocol used is the same as the one described in example 3.
Results of 3 independent experiments are presented in figure 44 and demonstrate that once purified 11E1 and 224G11 displayed no agonist effect when added alone to A549 cells and respectively 87 and 75 % antagonist effect in presence of VIGF.
As expected 5D5 Mab introduced as an agonist positive control showed a significant (58 %) agonist effect when added alone and only a moderate antagonist effect (39 %) in presence of HGF. Regarding to EC50 calculations, both 11E1 and 224611 had nanomolar ICsos.
Example 19: In vivo combination of 224611 and Navelbine on NCI-11441 xenograft model NCI-H441 cells from ATCC were routinely cultured in RPMI 1640 medium, 10 % FCS. I % L-Glutamine. Cells were split two days before engraftment being in exponential phase of growth. Ten million NCI-H441 cells were engrafted to Athymic nude mice. Five days after implantation, tumors were measurable and animals were divided into groups of 6 mice with comparable tumor size. Mice were treated i.p. either with a loading dose of 2 mg of 224611 Mab /mouse and then twice a week with 1 mg of antibody/mouse until Day 38 or with 3 injections of Navelbine (D5, D12, D19) at 8 mg/kg. A third group administered with the combine treatment was also included.
Navelbine was given by i.p. injections. Tumor volume was measured twice a week and calculated by the formula: ic/6 X length X width X height and animal weights were monitored every day over the period of Navelbine treatment. Statistical analysis was performed at each measured time using either a t-test or a Mann-Whitney test.
In this experiment, the average tumor volume of single modality treated groups is reduced by 72 %, 76 % and 99.8 % for 224011, Navelbine and Navelbine1-0 + 224011 respectively at day 41 post first injection. At day 41, the combined therapy improved significantly tumor growth compared to single therapy treatments (p_0.041 compared to Navelbine alone and pA.002 compared to 224G11 alone on day 41), 4 out of 6 mice being without tumors in the combined therapy group. Results are represented in figure 45.
These results were confirmed 50 days after the end of treatments (D88) where 66 % of mice receiving the combined treatment remained free of tumors.
Example 20: In vivo combination of 224G11 and Doxorubicine on NCI-H441 xenograft model NCI-H441 cells from ATCC were routinely cultured in RPMI 1640 medium, 10 % FCS, 1 % L-Glutamine. Cells were split two days before engraftment being in exponential phase of growth. Ten million NCI-H441 cells were engrafted to Athymic nude mice. Five days after implantation, tumors were measurable and animals were divided into groups of 6 mice with comparable tumor size. Mice were treated i.p. either with a loading dose of 2 mg of 224G11 Mab/mouse and then twice a week with 1 mg of antibody/mouse or with 4 injections of Doxorubicin (D5, D12, D19, D26) at 5 mg/kg. A
third group administered with the combine treatment was also included.
Doxorubicin was given by i.v. injections. Tumor volume was measured twice a week and calculated by the formula: 7c/6 X length X width X height and animal weights were monitored every day over the period of Doxorubicin treatment. Statistical analysis was performed at each measured time using either a t-test or a Mann-Whitney test. Both single therapies and combined treatment displayed significant anti-tumor activity compared to the control group (17t0.002 from Dll to D39). Results are represented in figure 46.
Combined treatment also demonstrates a significant anti-tumour growth activity compared to single modality treatment between Dll and D39 indicating that there is a benefit to combine Doxorubicin to an anti-c-Met treatment.
Example 21: In vivo combination of 224C11 and Docetaxcl on NCI-H441 xenograft model NCI-H441 cells from ATCC were routinely cultured in RPMI 1640 medium, 10 % FCS, 1 % L-Glutamine. Cells were split two days before engraftment being in exponential phase of growth. Nine million NCI-H441 cells were engrafted to Athymic nude mice. Five days after implantation, tumors were measurable and animals were divided into groups of 6 mice with comparable tumor size. Mice were treated i.p. either 5 with a loading dose of 2 mg of 2240'11 Mab/mouse and then twice a week with 1 mg of antibody/mouse or with 4 injections of Docetaxel (D5, D12, D19, D26) at 7.5 mg/kg. A
third group administered with the combine treatment was also included.
Docetaxcl was given by i.p. injections. Tumor volume was measured twice a week and calculated by the formula: R/6 X length X width X height and animal weights were monitored every 10 day over the period of Docetaxel treatment. Statistical analysis was performed at each measured time using either a t-test or a Mann-Whitney test. Both single therapies and combined treatment displayed significant anti-tumor activity compared to the control group (11Ø002 from Dll to D35). Results are represented in figure 47.
Combined treatment also demonstrated a significant anti-tumour growth activity 15 compared to single modality treatment between D18 and D35 indicating that there is a benefit to combine Docetaxel to an anti-c-Met treatment.
Example 22: In vivo combination of 224G11 and Temozolomide on U87MG
xenograft model 20 U87-MG cells from ATCC were routinely cultured in DMEM medium, 10 %
FCS, 1 % L-Glutamine. Cells were split two days before engraftment being in exponential phase of growth. Five million U87-MG cells were engrafted to Athymic nude mice. Nineteen days after implantation, tumors were measurable and animals were divided into groups of 6 mice with comparable tumor size. Mice were treated i.p. either 25 with a loading dose of 2 mg of 224G11 Mab/mouse and then twice a week with 1 mg of antibody/mouse or with 3 injections of Temozolomide (D19, D26, D33) at 5 mg/kg. A
third group administered with the combine treatment was also included.
Temozolomide was given by i.p. injections. Tumor volume was measured twice a week and calculated by the formula: n/6 X length X width X height and animal weights were monitored 30 every day over the period of Temozolomide treatment. Statistical analysis was performed at each measured time using either a t-test or a Mann-Whitney test.
Both single therapies and combined treatment displayed significant anti-tumor activity compared to the control group (p_0.002 from D22 to D32 (where control mice were euthanized for ethical reasons)). Results are represented in figure 48.
Combined treatment also demonstrate a significant anti-tumour growth activity compared to single modality treatments (10.002 from day 22 to day 43 (where control mice were cuthanized for ethical reasons) for Temozolomide and from day 29 to day 53 (last day of treatment) for 224G11. Taken together, these data indicate that there is a benefit to combine Temozolomide to an anti-c-Met treatment.
Example 23: Spheroid formation 1 0 As already described in Example 9 for other Mabs, we evaluate the ability of 224G11 Mab to inhibit in vitro tumor growth in the 1187-MG spheroid model. For that purpose, U87-MG cells grown as a monolayer were detached with trypsine-EDTA
and resuspended into complete cell culture media. Spheroids were initiated by inoculating 625 cells into single wells of round bottom, 96 plates in DMEM-2.5 % FCS. To prohibit cell adhesion to a substratum, the plates were pre-coated with polyHEMA in 95 %
ethanol and air dried at room temperature. The plates were incubated under standard cell culture conditions at 37 C, 5 % CO2 in humidified incubators. Purified monoclonal antibodies (10 lag/m1) were added after 4 and 10 days of spheroid culture.
Spheroids were kept in culture for 17 days. Then, spheroid growth was monitored by measuring the area of spheroids using automeasure module of axiovision software. Area was expressed in tm2. 8-16 spheroids were evaluated for each condition. Spheroid size was measured before addition of antibodies, after 10 days of culture and after 17 days of culture.
In those conditions, homogeneous spheroids were obtained and no statistical difference was observed before addition of antibodies (Figure 49A).
As illustrated in figures 49B-49D, isotype control, 9G4 did not affected growth of spheroids after 10 or 17 days of culture. While addition of 5D5 had no major effect on spheroid size, addition of either 224G11 and 11E1 markedly inhibited tumor growth.
Example 24: /n vitro activity of chimeric and humanized forms of 224G11 in the phospho-c-Met assay in order to compare the in vitro efficacy of murine, chimeric and humanized forms in a functional assay, culture supernatants resulting from 224G11 hybridoma, and HEK293 transfected cells were dosed and tested as described in Example 3. Data summarized in figure 50 showed the expected results for the unpurified murin antibody as already described in figure 6B. Both chimeric and humanized unpurified antibodies displayed a comparable activity either when added alone (figure 50A) or when incubated in presence of HGF (figure 50B).
Example 25: Determination of affinity constants (KD) of anti-c-Met antibodies by Biacore analysis The binding affinity of purified 11E1 and 224G11 antibodies was investigated by BlAcore X using recombinant e-Met-Extra-Cellular Domain (ECD) fused to an human lgG1 Fe domain (R&D Systems) as antigen (MW = 129 klla). As both c-Met-Fe fusion proteins and antibodies are bivalent compounds, Fab fragments of mAbs and 224G11 (MW = 50 klla) were generated by papain cleavage, purified and used in this assay to avoid interference with avidity parameter. For the assay, an anti-Tag histidine capture antibody was coated on CM5 sensorchips. The running buffer was HBS-EP, the flow rate was 30 gl/min and the test was performed at 25 C.
Soluble c-Met (ECD_M1)2-Fc-(HHHHHH)2 antigen was captured on the sensorchip (around 270 RU), and the antibodies to be tested were used in solution as analytes. The sensorship was regenerated using Glycine, HO pH 1.5 buffer on both floweells for half a minute.
Figure 51 illustrates the principle of this analysis. The resulting kinetic parameters are summarized in the following table 4. They indicate that both 11E1 and 224G11 anti-c-Met antibodies bind the recombinant c-Met-Fe fusion protein with comparable affinities ranging about 40 pM.
Table 4 K0n1 K0.1 X 1W6 X 1 Half-Life [1/M.s1 [1/M.s1 [hi [PK

1.13 + 0.01 4.68 0.001 4.1 41.4 0.5 Fab 2.04 0.01 7.79 + 0.40 2.5 34.8 1.9 Fab Example 26: In vivo activity of 224G11 on MDA-MB-231 cells co-implanted with MRCS cells as human HGF source on Athymic nude mice MDA-MB-213 and MRCS cells from ATCC were both cultured in DMEM
medium, 10 % FCS, 1 % L-Glutarnine. Cells were split two days before engraftment being in exponential phase of growth. Five million MDA-MB-231 cells and 500 IVIRC5 cells were co-injected s.c. to Athymic nude mice. Twelve days after implantation, tumors were measurable and animals were divided into groups of 6 mice with comparable tumor size. Mice were treated i.p. either with a loading dose of 2 mg of 224G11 Mab/mouse and then twice a week with 1 mg of antibody/mouse. Tumor volume was measured twice a week and calculated by the formula: Tr/6 X length X
width X height.
Results described in figure 52 showed a significant difference in median tumors growth of mice treated with 224G11 compared to the one of the control group.
Example 27: Complementary elements on humanization of antibodies 227H1, 11E1 and 224G11 General procedure Humanization of the anti-cMet antibodies were perfon-ned independently for each chain and sequentially, regarding to the analysed amino acids in each variable domain. The humanization process was evaluated in a first attempt in an ELISA-based binding assay to recombinant Fc-cMct; binding activities the humanized antibodies being compared to the recombinant chimeric antibody. In a second attempt, anti-cMet humanized antibodies were evaluated for their abilities to displace the Fc-cMet binding onto plastic-coated recombinant HGF; this competition assay allowing the direct comparison of murine, chimeric and humanized versions of the anti-Wet antibodies.
In Figures 53 and 54 are examplified the typical anti-cMct binding activities of 227H1, 11E1 and 224G11 murinc monoclonal antibodies.
Figure 53 shows anti-cMct direct binding activities of detected purified murinc antibodies. In this assay, murine monoclonal anti-cMet antibodies display different but still dose-dependent anti-eMet binding activities.
Figure 54 shows the HGF-cMet binding competition activities of purified murine antibodies. The competition assay reveals reliable differences between these anti-cMet monoclonal antibodies with a moderate, not full but reliable competitive activity for 11E1 monoclonal antibody whereas murine 224011 and 227H1 display similar pattern of competitive activities with a 100% of maximum of HGF
binding displacement at high antibody concentration. The 224011 monoclonal antibody showing the best IC50 value.
It is worthnoting that the direct binding activities of the murine antibodies do not reflect their intrinsic H0E-binding competitive properties.
These two assays were used to characterize the recombinant chimeric and humanized versions of the murine anti-cMet antibodies. To this end, briefly, anti-eMet variable domains, either murine or humanized, were cloned into 1,0NzA's pCONplus expression vectors series and recombinant IgGI/K-derived antibodies were expressed in CHO cells. Expression culture supernatants were concentrated and extensively dialysed against PBS and then dosed for expressed antibodies concentrations and directly used to assess corresponding anti-cMet binding activities. Both direct binding and HOE-competition assays were assessed to better characterize recombinant chimeric or humanized versions.
Example 27-1: Humanization of 227111 heavy chain variable domain In order to identify the best human candidate for the CDR grafting, the human germline gene displaying the best identity with the 227111 VH murine sequence has been searched. With the help of the 1MGT database, human 10HV1-2*02 V germline gene and human 10HJ4*01 J germline gene have thus been selected as receiving human sequences for the murine 227111 VH CDRs.
Figure 55 represents an amino acid alignment of the murine 227H1 VH domain with the selected human framework. In the human FR lane, only the amino acid that was found different from the 227111 murine VH domain is depicted. HZ3VH (SR.?
ID
N0:81), HZ2VH (SEQ ID NO:82) and HZ! VH (SEQ ID NO:83) lanes correspond to implemented humanized versions of the 227H1 VH domain with above ("changed in"

lane) mentioned mutations clearly identified. The number under each proposed mutation corresponds to the rank at which said mutation will be done.
In a first serie of experiments, we constructed and analysed the anti-cMet binding activities of the three first humanized versions of the 227H1 murine VH domain when expressed in combination with the 227H1 chimeric light chain. Results obtained from the anti-cMet direct binding assay are shown in Figure 56. In this experiment, no differences in the binding capabilities of the tested 227H1-derived chimeric or partially humanized recombinant antibodies were observed. At this point, 26 out of the 32 amino 5 acids that were found different between the murine 227H1 VH domain and the selected human framework have been analysed and found not relevant for anti-cMet binding activity of the 227H1 humanized VII domain, when combined with the chimcric light chain.
In conjonction with a site-directed rnutagenesis analysis of the last six murine 10 residues in the HZ1VH humanized version of the 227E11 VII domain, we constructed an original HZ4VH full-IMGT humanized >> version and tested its anti-cMet binding properties. Results are given in Figure 57 for the direct binding assay and in Figure 58 for the HGF binding competition assay. It is worthnoting that both the recombinant chimeric and humanized 227H1 versions display a better competitive activity than the 15 parental murinc antibody.
Nevertheless, given the experimental data obtained regarding the anti-cMet binding properties of the "full-IMGT" humanized 227H1 VH domain, the resulting amino acid sequence depicted in Figure 59 was selected and a bioinformatic analysis was then performed to evaluate the humaness level of the so-called 227H1-HZ VH
20 humanized variable domain.
To this end a simple comparison of the frameworks sequences to human database was performed using the IMGT tools. Given the humanization level that we reached during this process, out of the 89 analysed amino acids corresponding to the framework residues, 89 were found reliable with a human origin. Only residues from the CDRs can 25 be found different, but if so there are different from the corresponding human germline gene, and are obviously at hypervariable positions. Based on the IMGT
numbering system and homology analysis tools, we first totally humanized an antibody variable domain of murine origin.
30 Example 27-2: 11E1 monoclonal antibody humanization I - Humanization of 11E1 heavy chain variable domain In order to identify the best human candidate for the CDR grafting, the human gennline gene displaying the best identity with the 11E1 VII murine sequence has been searched. With the help of the IMGT database, human IGHV1-46*01 V germline gene and human IGHJ4*03 J germline gene have thus been selected as receiving human sequences for the murine 11E1 VH CDRs.
Figure 60 represents an amino acid alignment of the murine 11E1 VI-I domain with the selected human framework. In the human FR lane, only the amino acid that was found different from the 11E1 murine VH domain is depicted. IIZ V113 (SEQ ID: NO
84). 11Z
VH2 (SEQ Ill: NO 85) and HZ VI-11 (SEQ ID: NO 86) lanes correspond to implemented humanized versions of the 11E1 VH domain with above ("changed in" lane) mentioned mutations clearly identified. The number under each proposed mutation corresponds to the rank at which said mutation will be done.
In a first serie of experiments, we constructed and analysed the anti-cMet binding activities of the three first humanized versions of' the 11E1 murine VII
domain when expressed in combination with the 11E1 chimeric light chain. Results obtained from the anti-cMet direct binding assay are shown in Figure 61. In this experiment, a similar binding capability of the tested 11E1-derived chimeric or partially humanized recombinant antibodies was observed. At this point, 19 out of the 24 amino acids that were found different between the murine 11E1 VH domain and the selected human framework have been analysed and found not relevant for anti-cMet binding activity of the 11E1 humanized V1-I domain, when combined with the chimeric light chain.
II - Humanization of II El light chain variable domain In order to identify the best human candidate for the CDR grafting, the human germline gene displaying the best identity with the 11E1 VL murine sequence has been searched.
With the help of the IMGT database, human IGKV3D-7*01 V germline gene and human IGKJ4*01 J germline gene have thus been selected as receiving human sequences for the murine 11E1 VL CDRs.
Figure 62 represents an amino acid alignment of the murine 11E1 VL domain with the selected human framework. In the human FR lane, only the amino acid that was found different from the 11E1 murinc VL domain is depicted. 117 V1,3 (SEQ ID: NO
87), I IZ V1,2 (SEQ ID: NO 88) and 11Z VE1 (SEQ ID: NO 89) lanes correspond to implemented humanized versions of the 11E1 VL domain with above ("changed in" lane) mentioned mutations clearly identified. The number under each proposed mutation corresponds to the rank at which said mutation will be done.
In a first serie of experiments, was constructed and the analysed the anti-cMet binding activities of the three first humanized versions of the 11E1 murine VL
domain when expressed in combination with the 11E1 chimeric heavy chain. Results obtained from the anti-eMet direct binding assay are shown in Figure 63. In this experiment, we observed similar binding capabilities of the tested 11E1-derived chimeric or partially humanized recombinant antibodies. At this point, 26 out of the 30 amino acids that were found different between the murine 11E1 VL domain and the selected human framework have been analysed and found not relevant for anti-cM at binding activity of the 11E1 humanized VL domain, when combined with the chimeric heavy chain.
III - Humanization of 11E1 antibody At this stage of the 11E1 monoclonal antibody humanization, the theoretical resulting humanized antibody sequence contains only five outside-CDRs residues coming from the parental murine VH domain and four outside-CDRs residues coming from the parental murine VL sequence (see Figure 60, lane HZ VH1 and Figure 62, lane HZ VL1). It has then be decided to immediately characterize the resulting combined heavy and light chain humanized version of the 11E1 antibody. Results are given in Figure 64 for the anti-eMet direct binding assay.
In this experiment, it has been observed similar binding capabilities for the tested 11E1-derived chimeric or humanized recombinant antibodies. Analysis of the HGF-binding competitive properties and site-directed mutagenesis analysis of the contribution of the nine left murine residues remaining to be performed independently or in combination in this selected VH1/VL1 "pre-humanized" version of the 11E1 monoclonal antibody.
Example 27-3: 224G11 monoclonal antibody humanization 1- Humanization of 224G11 heavy chain variable domain In order to identify the best human candidate for the CDR grafting, the human germline gene displaying the best identity with the 224G11 VH murine sequence has been searched.
Regarding the high sequence homology between the 224G11 and the 227H1 VH
domains sequences, and as confirmed by the use of the IMGT database tools, the same human IGI1V1-2*02 V germline gene and human 10! IJ4*01 J germline gene have thus been selected as receiving human sequences for the murine 224011 VH CDRs.
Based on this high homology, it has been decided to directly transfert the humanization informations gained from the 227H1 VH domain humanization (see Example 27) and we then designed a "full-IMGT" humanized version (SEQ ID: NO
90) as depicted in Figure 65 which represents an amino acid alignment of the murine 227111 and 224011 VH domains with the selected human framework. In the human FR lane, only the amino acid that was found different from the 224011 murine VH domain is depicted. HZVHO lane corresponds to full-IMGT humanized version of the VH domain as obtained for the "full-1MGT" 227111-1-IZVII domain.
The full-1MGT humanized version of the 224011 murine VH domain has then been constructed and its anti-cMet binding activities were analysed, when expressed in combination with the 224G11 chimeric light chain. Results obtained from the anti-eMet direct binding assay are shown in Figure 66 while Figure 67 illustrates the 1 5 HGF binding competition assay. Given the experimental data obtained regarding the anti-cMet binding properties of the "full-IMOT" humanized 224011 VH domain, the resulting amino acid sequence as depicted in Figure 65 was selected and a bioinformatic analysis was then performed to evaluate the humaness level of the so-called 224G11-HZ VHO domain.
Given the humanization strategy applied here. it has to be referred to the Example 27 for the humaness analysis of the 224011 HZ V1-10 sequence. As described for the 227111 VH domain humanization, we confirm the reliability of the 'MGT
numbering system and homology analysis tools, and also demonstrate the possibility of transferring the humanization strategy between antibodies under the limits of their 2 5 intrinsec homology.
11 - Humanization of 224G11 light chain variable domain In order to identify the best human candidate for the CDR grafting, the human germline gene displaying the best identity with the 224011 VL murine sequence has been searched. With the help of the IMGT database analysis tools, two possible receiving human V regions for the murine 224G11 Vi.. CDRs were identified.
Thus, two humanization strategies have been planed for the 224011 VL domain. 'file first corresponds to an initial trial for a human framework with a shorter CDR' length (IGKV3-11*01), the second with a longer CDR1 length (IGKV4-1*01).
Figure 68 represents an amino acid alignment of the murine 224011 VL domain with the two selected human frameworks. In the both shorter and longer Hu-FR
FR
lanes, only the amino acid that was found different from the 224011 murine VL
domain is depicted. HZ VI,3 (SEQ NO 91) and HZ VL6 (SEQ ID: NO 92) lanes correspond to basic humanized versions of the 224011 VL domain with above ("rank"
lane) mentioned mutations clearly identified. The number under each proposed mutation corresponds to the rank at which said mutation will be done whenever the basic "shorter" or "longer" CDR1-framework will be selected.
In a first set of experiments, the two basic humanized versions of the 224011 murine VL domain were constructed and their anti-cMel binding activities were analysed, when expressed in combination with the 224011 chimeric heavy chain.
Results obtained from the anti-cMet direct binding assay are shown in Figure 69. In this experiment, a similar anti-cMet binding activity was observed for the chimeric and HZ
VL6 ( longer-CDR1 ) version whereas almost no binding was detected for the 1-V1.3 ( shorter-CDR1) recombinant 224011-derived antibody.
hi a second set of experiments, we constructed and analysed the anti-cMet binding activities of the implemented humanized versions of the HZ VL6-derived 224011 VI-1 domain when expressed in combination with the 224011 chimeric heavy chain. Two additional humanized form was analysed; in the HZ VI.5 version the seven residues from the third group (rank 3) are humanized and in the 11/ V1,4 version the four left residues from the first group (rank 1 residues) only remained murine. Results obtained from the anti-cMet direct binding assay are shown in Figure 70. In this experiment, no differences in the binding capabilities of the tested 224011-derived 2 5 chimeric or partially humanized recombinant antibodies were observed. At this point.
18 out of the 22 amino acids that were found different between the murinc 224011 VI.
domain and the selected longer-CDR I human framework have been analysed and found not relevant for anti-cMet binding activity of the 224011 humanized VL
domain, when combined with the chimeric heavy chain.
It has then be tested the HZ VL4 humanized version of the 224011 VI, domain in the HGF binding competition assay. As shown in Figure 71, the results obtained demonstrate the similar competitive activity of murine and recombinant chimeric and HZ VL4 humanized 224G1I-derived antibodies.
At this stage of the 224G11 VL domain humanization, the resulting sequence contains only four outside-CDRs residues coming from the murine parental sequence.
As shown in Figure 72, these four -labelled residues are L4, M39, H40 and R84.
5 Based on the IMGT numbering system and homology analysis tools, we demonstrated that human framework displaying structural differences in term of CDR
length may still be suitable in a humanization process. It has then been decided to characterize the resulting heavy and light chain humanized version of the antibody. Site-directed mutagenesis analysis of the contribution of the remaining four 10 murine residues being then performed when expressed in combination with the VHO
humanized version of the heavy chain.
Ill - Humanization of 224G11 antibody In a first scrie of experiments, we constructed and analysed the anti-cMet binding activities of the fully humanized version of the 224G11 antibody. This 15 recombinant version encompass both VHO and VL4 humanized VH and VL domains respectively. Results obtained from the anti-cMet direct binding assay are shown in Figure 73. In this experiment, the fully human 224G11 anti-cMet binding activity was found similar to that of single-chain >> humanized and chimeric recombinant versions.
20 It has then been tested the fully humanized version of the 224G11 VL domain in the HGF binding competition assay. The results obtained as shown in Figure 74 demonstrate the similar competitive activity of parental murine and recombinant chimeric and fully humanized 224G11-derived antibodies.
At this stage of the 224G11 antibody humanization, the resulting sequence 25 contains only four outside-CDRs residues coming from the murine parental light chain variable domain sequence. We then analysed site-directed mutagenesis single variants of the VL4 humanized VL domain when expressed in combination with the VHO
humanized version of the heavy chain. As examplified in Figure 75 for the direct binding assay we identified potential relevant residues among the four tested, being 30 M39 and H40.
It has been decided to analyse multiple mutants of the HZ VL4 humanized 224G11 VL domain when expressed in combination with the HZ VHO humanized 224G11 VH domain. As shown in Figure 76 for the direct binding assay and in Figure 77 for the HGF binding competition assay, multiple amino acids mutants of the domain were analysed to identify the best humanized combination. Based on the single mutants analysis, it has been focused on double and triple mutants that may exhibits the best anti-cMet activities. The VHO/VL4-2x mutant correspond to the HZ VHO

humanized VH domain expressed with the HZ VL4 224011 humanized VL domain with the double mutation L4M/R84G. The VHO/VL4-3x mutant correspond to the HZ
VHO 224011 humanized VH domain expressed with the HZ VL4 224011 humanized VL domain with the triple mutation L4M/M39L/R84G.
Given the experimental data obtained regarding the anti-cMet binding properties of the fully humanized 224G11 antibody the bioinformatic analysis of both heavy and light chain variable domains sequences was then performed to evaluate the humancss level of the VHO/VL4-2x and VHO/VL4-3x best humanized versions. It has been previously demonstrated the "full-1MGT" humanization of the VHO

VH domain. Regarding the humancss level of the VL4-2x and -3x 224011 humanized VL domain versions, they only contain murine residues 1V139 and/or H40. These two potential key residues are located at the end of the CDR1, M39 being the N-terminal CDR anchor. Given the CDR length problem that we faced during the 224011 VL
domain humanization, and considering those positions as part of the Kabat definition of the VL CDR1, the humaness level of the fully humanized 224011 antibody should display a strongly reduced immunogenicity due to the minimal conserved murine residues.

Claims (45)

92

1. An isolated antibody, or one of its functional divalent fragments capable of binding specifically to c-Met, characterized in that said isolated antibody or said functional divalent fragment comprises a heavy chain comprising CDR-H1, CDR-H2 and CDR-H3 comprising respectively the amino acid sequence SEQ ID Nos. 7, 8 and 9; and a light chain comprising CDR-L1, CDR-L2 and CDR-L3 comprising respectively the amino acid sequence SEQ ID Nos. 15, 16 and 17.

2. The antibody of claim 1, or one of its functional divalent fragments, characterized in that it comprises a heavy chain comprising the amino acid sequence SEQ ID NO:
20 and a light chain comprising the amino acid sequence SEQ ID NO: 23.

3. An antibody secreted by the murine hybridoma deposited at the CNCM, Institut Pasteur, Paris, on March 14, 2007 under the number I-3786.

4. The antibody of claim 1 or 2, or one of its functional divalent fragments, characterized in that it consists in a monoclonal antibody.

5. The antibody of claim 4, or one of its functional divalent fragments, characterized in that said antibody is a chimeric antibody and further comprises light chain and heavy chain constant regions derived from an antibody of a species heterologous to the mouse.

6. The chimeric antibody of claim 5, or one of its functional divalent fragments, characterized in that said heterologous species is human.

7. The chimeric antibody of claim 6, or one of its functional divalent fragments, characterized in that the light chain and heavy chain constant regions derived from a human antibody are respectively for the light chain the kappa region and for the heavy chain the gamma-1, gamma-2 or gamma-4 region.

8. The antibody of any one of claims 1 to 7, or one of its functional divalent fragments, characterized in that it is capable to inhibit ligand-dependent and ligand independent activation of c-Met.

9. The antibody of claim 8, or one of its functional divalent fragments, characterized in that it inhibits the c-Met dimerization.

10. The antibody of claim 9, or one of its functional divalent fragments, characterized in that it inhibits at least 50 % of tumoral cell proliferation for at least one tumor type.

11. The antibody of any one of claims 1 to 10, wherein said antibody is coupled chemically to a cytotoxic agent.

12. The antibody of claim 11, wherein said cytotoxic agent is an alkylating agent, anti-metabolite, anti-tumor antibiotic, mitotic inhibitor, chromatin function inhibitor, anti-angiogenesis agent, anti-estrogenic agent, anti-androgen agent or immunomodulator.

13. The antibody of claim 11 or 12, wherein said cytotoxic agent is a mitotic inhibitor.

14. A murine hybridoma capable of secreting an antibody as defined in claim 2, characterized in that said murine hybridoma is the murine hybridoma deposited at the CNCM, Institut Pasteur, Paris, on March 14, 2007 under the number 1-3786.

15. An isolated nucleic acid coding for the antibody, or one of its functional divalent fragments, as defined in any one of claims 1 to 10.

16. The nucleic acid of claim 15, wherein said nucleic acid is:
a) a nucleic acid comprising a DNA sequence comprising the sequences SEQ ID
No. 30, SEQ ID No. 31, SEQ ID No. 32 and the sequences SEQ ID No. 38, SEQ ID No. 39 and SEQ ID No. 40; or b) a nucleic acid comprising a DNA sequence comprising the sequences SEQ ID
NO: 43 and SEQ ID NO: 46.

17. A vector comprising the nucleic acid as defined in claim 15 or 16.

18. A host cell comprising the vector as defined in claim 17.

19. A process for production of the antibody, or of one of its functional 25 divalent fragments, as defined in any one of claims 1 to 13, characterized in that it comprises the following stages:

a) culture in a medium and appropriate culture conditions of the host cell as defined in claim 18; and b) the recovery of said antibody, or of one of its functional divalent fragments, thus produced starting from the culture medium or said cultured cells.

20. A pharmaceutical composition comprising (a) a compound consisting of the antibody, or of one of its functional divalent fragments, (i) as defined in any one of 5 claims 1 to 13; or (ii) obtained by the process of claim 19, and (b) a pharmaceutically acceptable carrier.

21. The pharmaceutical composition of claim 20, further comprising an anti-tumoral antibody.

22. The pharmaceutical composition of claim 21, characterized in that said second anti-tumoral antibody is selected from anti-IGF-1R, anti-EGFR, anti-HER/neu, anti-VEGFR
or anti-VEGF antibodies.

23. The pharmaceutical composition of claim 22, characterized in that said second anti-tumoral antibody is Erbitux.

24. The pharmaceutical composition of any one of claims 20 to 23, further comprising a cytotoxic/cytostatic agent.

25. The pharmaceutical composition of claim 24, wherein said cytotoxic/cytostatic agent is selected in the group consisting of alkylating agents, anti-metabolites, anti-tumor antibiotics, mitotic inhibitors, chromatin function inhibitors, anti-angiogenesis agents, anti-estrogens, anti-androgens or immunomodutators.

26. The pharmaceutical composition of claim 25, wherein said cytotoxic/cytostatic agent is gefitinib or erlotinib.

27. The pharmaceutical composition of claim 25, wherein said cytotoxic agent is a mitotic inhibitor.

28. The pharmaceutical composition of any one of claims 21 to 28, characterized in that one, at least, of said antibodies, or a divalent functional fragment or derivative thereof, is conjugated with i) a cell toxin and/or, ii) a radioelement, or iii) both a cell toxin and a radio-element.

29. The pharmaceutical composition of claim 20, further comprising a cytotoxic/cytostatic agent.

30. The use of the antibody, or one of its functional divalent fragments (i) as defined in any one of claims 1 to 13, or (ii) obtained by the process as defined in claim 19, for the preparation of a medicament for inhibiting the growth, the proliferation, or the growth and proliferation of c-Met-expressing tumor cells.

31. The use of (a) the antibody, or one of its functional divalent fragments (i) as defined in any one of claims 1 to 13, or (ii) obtained by the process as defined in claim 19; or (b) of claim 30, wherein the medicament is for preventing or treating a cancer expressing c-Met.

32. The use of the pharmaceutical composition as defined in any one of claims 20 to 29, for the preparation of a medicament for inhibiting the growth, the proliferation, or the growth and proliferation of c-Met-expressing tumor cells.

33. The use of the pharmaceutical composition of claim 29, for the preparation of a medicament for inhibiting the growth, the proliferation, or the growth and proliferation of c-Met-expressing tumor cells, wherein the antibody and cytotoxic/cytostatic agent are administered simultaneously, separately, or sequentially.

34. The use of (a) the pharmaceutical composition as defined in any one of claims 20 to 29; or (b) of any one of claims 32 or 33, wherein the medicament is for preventing or treating a cancer expressing c-Met.

35. The use of any one of claims 30 to 34, characterized in that said cancer expressing c-Met is prostate cancer, osteosarcoma, lung cancer, breast cancer, endometrial cancer, glioblastoma or colon cancer.

36. The use of any one of claims 30 to 34, characterized in that said cancer expressing c-Met is prostate cancer.

37. The use of any one of claims 30 to 34, characterized in that said cancer expressing c-Met is osteosarcoma.

38. The use of any one of claims 30 to 34, characterized in that said cancer expressing c-Met is lung cancer.

39. The use of any one of claims 30 to 34, characterized in that said cancer expressing c-Met is breast cancer.

40. The use of any one of claims 30 to 34, characterized in that said cancer expressing c-Met is endometrial cancer.

41. The use of any one of claims 30 to 34, characterized in that said cancer expressing c-Met is glioblastoma.

42. The use of any one of claims 30 to 34, characterized in that said cancer expressing c-Met is colon cancer.

43. The use of any one of claims 30 to 43, characterized in that said cancer expressing c-Met is a hepatocyte growth factor (HGF) dependent Met-activation related cancer or a HGF independent Met-activation related cancer.

44. A method of in vitro diagnosis of cancers induced by an overexpression of c-Met receptor starting from a biological sample in which the abnormal presence of c-Met receptor is suspected, characterized in that said method comprises:
a) a step wherein said biological sample is contacted with the antibody, or one of its functional divalent fragments (i) as defined in any one of claims 1 to 13, or (ii) obtained by the process as defined in claim 19; and b) a step wherein the overexpression of the c-Met receptor is determined by the demonstration of the c-Met/antibody complex formed in step a).

45. The method of claim 44, wherein the antibody is labeled.